Each one is complete and ready to use. You open a list, click once, and every item is added to your own list in Valedo.

Start with dinner

Say you decide to cook Ukrainian borsch on Saturday.

Normally, you find a recipe, read through it, then start typing ingredients into your shopping list one by one.

Beef on the bone. Beets. Cabbage. Potatoes. Carrots. Onion. Tomato paste. Garlic. Bay leaf. Dill. Sour cream. Vinegar.

In Valedo, you open the borsch recipe and add it to your list. Every ingredient, including the amounts, goes directly onto the Grocery shopping list you already share with your wife.

She sees it on her phone.

You check what is already in the refrigerator and uncheck those items. What remains is what you need to buy.

Nobody copied anything. Nobody had to remember anything. Nothing got lost between the recipe and the store.

Then the lists start stacking

This is where the feature became more interesting to me.

Suppose you are having a Mexican BBQ on Sunday.

You add the carne asada recipe with the full marinade. Then salsa. Rice and beans. Then a grill supplies list with charcoal, foil pans, tongs, napkins, and ice.

All of it can go onto the same list as Saturday’s borsch.

Now one shopping trip covers the weekend.

You can share the list with your brother and give him check-only access. He picks up the ice and drinks, checks them off, and cannot accidentally delete half the menu.

A recipe stopped being something you read.

It became part of an actual plan.

Food is just the obvious example

You are moving in three weeks.

Add the moving checklist and you get boxes, tape, markers, bubble wrap, utility transfers, address changes, documents to update, and supplies for cleaning the old place.

Then add the new apartment setup list.

Two clicks and you have the starting point for the next three weeks.

Or a hurricane is approaching Florida.

Every season, people make another trip to the store and try to remember what they forgot during the last storm.

Water. Batteries. Flashlights. Cash. Medication refills. Power banks. Chargers. Important documents in a waterproof bag. Gas in the car.

Add the hurricane preparation list, share it with the family, and split the errands.

School starts Monday? Add the school list.

A puppy comes home Friday? Add the first-week puppy list with the crate, bowls, food, leash, chew toys, waste bags, and the vet appointment you should probably book before the puppy arrives.

Going camping for the first time? There is a list for that too.

The best lists may be for things you have never done

This is the part I find most useful.

If you have cooked borsch fifty times, you probably do not need anyone to tell you to buy beets.

But what if you are bringing home your first newborn?

What if you just got your first dog?

What if you are taking your first camping trip, moving into your first apartment, preparing for your first hurricane, or organizing a birthday party for twenty children?

The problem is not remembering your list.

You do not know what should be on the list yet.

A blank screen is terrible at solving that problem.

It basically says: Here is an empty box. Good luck.

A ready-made list gives you a starting point based on what the task actually requires.

You do not have to use everything.

Maybe the camping list has 20 items and you keep 12. It still did its job. You reviewed 20 things and decided what mattered instead of trying to invent those 12 from memory.

That is a very different way to use a list app.

Why we put all 2,050 lists on the web

I made two decisions before publishing them.

The first was to make the lists public.

You do not need a Valedo account to read them.

If someone searches for a borsch shopping list, first apartment checklist, hurricane preparation list, puppy supplies, or what to pack for a three-day camping trip, I want them to be able to find the list and use it.

If they want to save it, customize it, combine it with other lists, share it with family, or check things off together, that is where Valedo becomes useful.

I do not think someone should create an account just to discover whether a packing list is worth looking at.

Show the value first.

And this is just the first batch

There are now 2,050 ready-made lists in Valedo, and we are already working on the next big batch.

The goal is simple: whatever you are planning, buying, cooking, organizing, preparing for, or trying for the first time, there should be a useful list you can start with instead of building one from zero.

And when there isn’t one, you can still create your own.

Valedo is free. Download the app, find a list you need, add it with one click, and make it yours.

If you try one of the 2,050 lists and find something missing, tell me. That is exactly how the next version gets better.

Green lid. The small one.

By the time I am standing in front of the shelf at Costco, that photo is eleven messages up, somewhere between a voice note, a photo of something else we need, and a completely unrelated family conversation.

I scroll.

I scroll some more.

At some point I give up and buy the one with the blue lid.

I come home.

Wrong one.

This is a very small problem. But it is the kind of small problem that happens all the time.

During the day I notice we are almost out of paper towels. I think, I need to remember to buy them.

Two hours later, that thought is gone.

My wife gives me a handwritten shopping list. I take a photo of it because obviously I am not going to carry a piece of paper around.

Then I forget that I took the photo.

Sometimes I put things in Notes. Sometimes they are in messages. Sometimes someone sends me a screenshot. Sometimes I just trust my memory, which is usually the worst option.

Travel is the same.

A few days before a trip, you suddenly start remembering things at random times.

Passport.

Phone charger.

Medicine.

Sunscreen.

Something for the kids.

The thing you absolutely must take with you and will somehow remember only when you are already twenty minutes from home.

None of this is a serious problem.

It is just unnecessarily annoying.

And that is basically why we built Valedo.

Valedo is a simple place for the lists we actually use in everyday life.

Shopping lists. Travel lists. Wishlists. Things to do. Things not to forget.

Nothing complicated.

Sometimes the list already exists

My daughter comes home from school with a supply list.

It is printed on paper, folded in half, and has twelve things we need to buy.

I have absolutely no desire to type twelve items into my phone.

So I take a photo.

Valedo turns it into a list I can actually use and check off while shopping.

The same works with a handwritten shopping list.

That sounds like a small feature until you realize how often someone gives you a list that already exists somewhere.

Why type it again?

The same goes for voice.

You are cooking and notice that the olive oil is almost finished.

Your hands are busy.

You just say, “olive oil,” and add it.

That is how I think these things should work.

The less effort it takes to remember something, the more likely you are to actually remember it.

Then there is family

Lists become even more useful when more than one person is involved.

Imagine we are preparing for a trip.

I create the list and share it with the family.

Now everybody can see what is already packed or purchased instead of asking:

“Did you get this?”

“Who is bringing that?”

“I thought you were buying it.”

“No, you said you were buying it.”

Anyone with a family probably knows this conversation.

The same thing happens with birthday parties, dinners, school events, vacations, and even a normal weekend.

One shared list is much easier than trying to manage everything inside a group chat.

And sometimes I want people to help with the list without completely changing it.

So some people can simply check things off, while others can add or edit.

Basically, you can share the list without giving everyone the pen.

We also keep rebuilding the same lists

A birthday party ends.

The shopping list is finished.

You bought plates, candles, drinks, snacks, decorations, and twenty other things.

Then next year comes around and you start from zero again.

Why?

In Valedo, completed lists can stay in your History.

Next year you open the old one, duplicate it, make a few changes, and you are done.

Same with camping.

Travel.

School supplies.

Moving.

Holiday dinners.

There are many things in life that feel new every time but are actually 80% the same list.

And sometimes you just need a starting point

The blank screen can be surprisingly annoying.

You know you are going camping, but what exactly should be on the list?

Tent.

Flashlight.

Bug spray.

And then seventeen things you will remember after you arrive.

So Valedo also has ready-made lists you can start from.

Camping, moving, newborn, travel, school, and other common situations.

You do not have to follow them.

They are just there to remind you of things you might otherwise forget.

Sometimes that is enough.

It also had to work everywhere

I wanted Valedo to work on the web as well as on phones.

That actually made building it slower, but for me it was important.

I might make a list while sitting at my computer and then open the same list later from my phone at the store.

Someone else might receive a shared list and simply want to open it in a browser.

Because once someone says, “I can’t open this,” you know what happens next.

Screenshot.

Group chat.

And we are back to looking for the green lid eleven messages up.

The whole idea is pretty simple

I do not think people need another complicated productivity system.

Most of the things we forget are very ordinary.

Buy milk.

Bring a charger.

Call someone.

Get something for school.

Pack sunscreen.

Pick up a birthday gift.

Remember the exact thing your wife asked you to buy instead of coming home proudly carrying the wrong version.

Valedo is simply a place to keep those things together.

It is free, the onboarding is simple, and you can start using it almost immediately.

So if any of the situations I described sound familiar, download Valedo and give it a try. (https://valedo.app)

Make a list. Share it with your family. Add things whenever you remember them.

Hopefully you forget fewer things, buy fewer wrong things, and spend less time standing in Costco scrolling through old messages.

And if nothing else, maybe you finally come home with the green lid.

It is also a topic I keep ending up in conversations about. Everyone has read something different, and nobody is working from the same set of facts.

So I decided to collect the data myself, and to collect it fresh. Asking an AI model is not enough here. Their training data has a cutoff, and this field moves. Papers from this year overturn papers from two years ago.

Tech stack: Hexomatic to scrape, Second Brain to import and index, Claude to analyze.

The process

Step one. Scrape the video content.

I pulled the top longevity podcasts and videos with their full transcripts. 285 videos, total runtime 170 hours.

Huberman, Attia, Sinclair, Rhonda Patrick, Stanfield, Topol, Valter Longo, and a long tail of smaller channels.

Step two. Scrape the articles.

Then I pulled 1,074 article URLs. Peer-reviewed papers, institutional pages, clinical trial registrations, and supplement vendor blogs.

Step three. Import into Second Brain.

Both datasets went in as structured tables with full text, source URLs, dates, channel names and runtimes. That made it queryable.

Step four. Analyze with Claude, then check every claim against the source.

Twenty-seven rounds of review and fact-checking over two days. Every number opened at the journal it came from.

What the data showed

Every source was counted for which compounds it mentions.

Taurine, acarbose, dasatinib and canagliflozin are researched and never discussed. Magnesium, multivitamin and ashwagandha are discussed and barely researched. The prescription drugs sit almost entirely on the article side. The supplements sit almost entirely on the video side.

The short version

No supplement in this data has been shown to slow aging. Eight things have real human evidence behind them.

1. Cardiovascular exercise. Dying less, of anything. Large cohorts rather than trials.

2. Resistance training. Strength, muscle mass, fewer falls. Trials.

3. Enough protein. Holding onto muscle after 50. Trials, with the target still argued over.

4. Creatine. Strength and lean mass. Trials. Memory, less reliably.

5. Omega-3. Fewer infections and falls. Trials. Fewer heart attacks, but only as a side measure.

6. Vitamin D, if you are deficient. Correcting deficiency. Trials. A survival signal in long trials, from subgroup analysis rather than a primary result.

7. A daily multivitamin. Cognition, by a small margin. Trials, not yet confirmed.

8. Sleep. Dying less at around seven hours. Cohorts rather than trials.

Every other supplement and drug here falls into one of four groups. It worked in mice and nowhere else. It moved a blood marker without moving an outcome. It came back mixed or negative when a trial was run in people. Or nobody has run one.

Part 1. What holds up in humans

Trials and cohorts in people, with real endpoints.

Exercise

In a study of more than 120,000 adults given treadmill tests, the unfit were several times more likely to die of any cause than the very fit, with no upper limit to the benefit. The authors put the danger of being unfit at or above smoking, diabetes, and high blood pressure.[1]

Attia and Huberman describe the same weekly structure: a few hours of easy aerobic work, one session hard enough to push VO2 max, three or four sessions of lifting.

In DO-HEALTH, a three-year trial in adults over 70, a home exercise program moved none of the four biological age tests.[2] Longevity products are commonly judged against those same measures.

Resistance training and protein

After 50, muscle mass and strength decline every year, with the lower body going roughly twice as fast as the upper. Once that loss becomes sarcopenia, fall rates in over-65s rise sharply.[3]

Two protein targets circulate. Attia argues the higher one, on the grounds that older muscle needs more protein to respond at all. The lower one comes from an analysis that found a ceiling above which extra protein added nothing. In that paper the ceiling was not statistically significant, and the uncertainty around it was wide enough to cover both targets.[4]

A Harvard and USC group looked at protein source rather than total, and found the highest intakes associated with more type 2 diabetes, entirely from animal protein. That is a cohort, not a trial.

Creatine

Creatine comes up in 50 videos and 38 articles. Combined with resistance training it beats training alone on lower-body strength and lean mass in older adults. Upper-body strength showed no significant gain. The effects are small and appear across trials.[3]

Muscle and brain need different doses, a distinction Darren Candow draws and most of the videos miss. The dose that saturates muscle is a quarter of what the memory trials used.

The main review of the memory studies drew a published letter arguing that double-counting had produced false positives.[5] A later review found effects on memory and attention but nothing on overall thinking ability.[6] Neither turned up in the videos.

Omega-3

That same trial gave its participants vitamin D, omega-3, a home exercise program, or combinations.[2] In a post hoc analysis of roughly a third of the participants, omega-3 alone slowed three of the four biological age tests, by a matter of months. Vitamin D and exercise moved nothing. Earlier results from the full trial were larger, with fewer infections and fewer falls.[2]

Four independent experts reviewed it. The omega-3 arm was small, the four tests gave different answers about the same participants, and nobody has data past three years.[7]

The largest fish oil trial in this dataset, in more than 25,000 people, set out to measure heart attack, stroke, and cardiac death together, and found no significant reduction. Heart attack counted on its own, which was a side measure, did come down.[8] Brad Stanfield quotes the second result without the first.

Vitamin D, if you are deficient

Three large trials, tens of thousands of people. In adults who already had enough, supplementing prevented no cancer, no heart attacks, no falls, no diabetes, and no bone loss.[9] A review of the whole literature found a survival benefit only in people already ill with something specific.[10] A 2025 review found no effect on heart attacks or strokes.[11]

One finding points the other way. In trials that ran longer than three years, people taking vitamin D died less often. That came from subgroup analysis, and the subgroups that benefited were on modest doses, were under 80, and started out deficient.[12]

A daily multivitamin

A three-year trial in adults over 65 compared cocoa extract, a multivitamin, both, and neither. Cocoa did nothing. The multivitamin improved cognition on all three measures.[13]

In a smaller follow-up group the overall cognition result could no longer rule out zero, though memory held.[14] The figure repeated in the videos, 60 percent slower cognitive aging, comes from the trial’s own results page, which says it needs confirming.[15]

Sleep

The mortality curve bottoms around seven hours, and the penalty for sleeping too much is steeper than for sleeping too little.[16] In a large cohort wearing objective trackers, how consistent your sleep timing was predicted mortality better than how long you slept.[17]

Deep sleep declines with age and tracks with Alzheimer’s risk. The mechanism usually given, that deep sleep flushes waste out of the brain, was contradicted by a 2024 mouse study that found clearance reduced during sleep rather than increased. Objections to that study’s methods have since been published, so the question is open.[18]

Stanfield’s melatonin argument is about dose. The trial that compared three doses found the physiological one restored sleep efficiency and brought blood levels back to normal. The pharmacological dose also improved sleep, but it caused hypothermia and left melatonin elevated into the following day.[19] The dose sold in most shops is about ten times the physiological one.

Part 2. Diet, lifestyle, and markers

Not supplements. The first is heavily evidenced and rarely discussed. The rest are heavily discussed.

ApoB

ApoB-carrying particles cause heart disease. That is the European Atherosclerosis Society consensus position, drawn from genetic, epidemiological and trial evidence, and genetic studies put ApoB ahead of LDL cholesterol as the main driver.[20]

Two things follow. Lab reference ranges are population percentiles rather than risk thresholds, so a number inside the printed range tells you that you are typical, not that you are safe. And because arterial damage accumulates from early adulthood, lowering ApoB earlier should count for more than lowering it late.

ApoB carries more evidence than anything else in this section, and appears in 11 videos and 1 article.

Fasting

78 videos and 45 articles, more than any non-supplement topic except protein. The general argument is that intermittent restriction makes cells perceive adversity and switch on repair.

The guest on Nikhil Kamath’s podcast pushes back: fasting is no better than a plain calorie deficit, and the Nobel Prize was awarded for work done in single-cell yeast. The rest of the sources disagree.

Sauna, cold, fiber, diet patterns

Sauna gets 26 videos and 1 article, cold 25 and 2, fiber 63 and 22, Mediterranean diet 35 and 9. On diet the sources agree that no particular one wins and the best is whichever a person can keep doing. On the rest there is almost nothing to check against.

Part 3. Works in mice, unproven in people

The National Institute on Aging funds a program that tests longevity drugs in mice, running the same protocol at three separate labs and pooling the results.[21]

Rapamycin

Rapamycin produced the largest effect the program has recorded, extending typical lifespan substantially in both sexes and still working when started in late middle age.[21]

Most of the human work has used related drugs rather than rapamycin itself. Those trials found the drugs tolerated and reported immune benefits, including a better flu vaccine response in older adults, but nothing on lifespan or healthy years.[22] It is a prescription immunosuppressant.

The other drugs that worked

None of the 285 videos mentions acarbose or canagliflozin, though both extended mouse lifespan by more than any supplement the videos discuss. Acarbose extends male lifespan more than rapamycin does, and does much less for females. 17-alpha estradiol and canagliflozin work in males and not females. Glycine works in both. Seven others work in males only.[21][23]

Of the compounds the program has shown to extend lifespan, most work only in males. Not one works only in females.[21]

Calcium alpha-ketoglutarate

Extended lifespan in middle-aged female mice and improved health in both sexes, with no significant side effects on long-term dosing.[24] The lifespan effect was modest and appeared in one sex only. The human study behind the marketing was not randomized and used a biological age test that has not been validated.

Part 4. Tried in people, came back mixed

Urolithin A

The pivotal trial reported stronger hamstrings and lower inflammation, with grip and quadriceps strength moving but not significantly. It missed the main thing it set out to measure. The corresponding author works at the company that sells it.[25] A later review adds that not everyone carries the gut bacteria needed to produce the compound.[26]

GlyNAC

A trial at Baylor gave twelve older adults GlyNAC and twelve a placebo for sixteen weeks. Antioxidant levels rose, damage markers fell, and walking speed improved.[27]

Nestlé then ran it in nearly ten times as many people and missed the main thing it set out to measure, except in one subgroup picked out after the fact.[28]

NAD boosters: NR and NMN

Both roughly double NAD levels in the blood, and neither beats the other. They work indirectly, through gut bacteria.[28] Short-term trials in older adults report minimal change in thinking, blood vessels, or muscle. Nicotinamide riboside was tested in the mouse program and did not extend lifespan.[23]

Senolytics: dasatinib plus quercetin

Senolytics are meant to clear out worn-out cells that stop dividing but never get removed.

In a trial at Mayo in postmenopausal women, markers of bone formation rose transiently and markers of bone breakdown did not move. The comparison across the whole group was not significant, and only the women who started with a high burden of those cells responded.[29]

The trial was designed with a fisetin arm too. It was dropped during COVID for recruitment and cost reasons, so there is no fisetin data.[29]

Part 5. Failed

Resveratrol

The mouse program tested it at two doses and neither extended lifespan.[23] The apparent mechanism turned out to depend on the assay. Resveratrol’s activation of the target enzyme only appeared when a tagged substrate was used to detect it. GSK bought Sirtris, the company built on the compound, in 2008 and closed its Cambridge operation in 2013.

David Sinclair, who co-founded Sirtris, still takes resveratrol. He points to a result in the supplemental data of the original mouse paper. It has not been replicated.

Metformin as a longevity drug

Metformin on its own did not extend lifespan in the mouse program.[23]

Sinclair takes it daily and calls it a longevity drug, citing an analysis of diabetics on metformin who outlived a non-diabetic comparison group. That is observational data from a treated patient population.

He also argues that the widely repeated claim about metformin blunting exercise adaptation came from a small difference presented on a compressed axis.

Fisetin

The mouse program dosed it both continuously and intermittently, at doses well above what a person would take. No lifespan extension in either sex, and no statistically significant reduction in the markers for worn-out cells. Without a change in those cells, the researchers could not test the mechanism the compound is sold on.[23]

Earlier work in a different mouse strain had reported both clearance and a lifespan gain. The program did not reproduce it, and the discrepancy is unresolved.

Taurine

In 2023, Vijay Yadav’s team published in Science that taurine falls with age across humans, mice, and monkeys, and that supplemented mice lived longer.[30] Sales followed.

In 2025 the same journal published the opposite. A group at the NIH followed the same individuals over time rather than comparing young people to old, and found taurine did not fall. It rose or stayed flat, and the gap between two people of the same age was larger than any change either went through.[31][32] Their summary was that there is no need to supplement if you eat reasonably.[33][34]

Yadav has not conceded and is running a controlled trial.[35]

Collagen peptides

Stanfield says collagen reduces wrinkles by about 8 percent, citing an analysis that pooled more than twenty trials.

That analysis says more. Taken together, collagen did improve hydration, elasticity, and wrinkles. Split by who paid for the trials, those without industry funding showed no effect on any of the three, and neither did the better-designed ones. The authors conclude there is currently no clinical evidence supporting collagen for skin aging.[36]

Cocoa, aspirin, methylene blue

Cocoa extract did nothing for thinking ability over three years.[13] Aspirin extended the lives of male mice in one trial, and the result did not reproduce. Methylene blue did nothing in either sex and was dropped.[23]

Part 6. Thin or absent

Pushed hard, thinly evidenced

Magnesium. 87 videos, 9 articles. The review of insomnia studies found people fell asleep somewhat faster, with no gain in total sleep time, and its authors said the literature is not good enough to make recommendations from.[37] A 2025 review found only a handful of controlled studies, most in people without sleep problems.[38]

TMG. Stanfield takes it. It lowers a blood marker linked to Alzheimer’s, but most trials show no strength gain, and nothing here shows it prevents dementia.

Psyllium. Stanfield takes it to feed gut bacteria, with reported gains in blood sugar and cholesterol.

Sulforaphane. Rhonda Patrick takes it to raise the body’s main antioxidant, and says the detoxification part is inferred rather than demonstrated.

Hormone replacement therapy. Argued in one review as something that slows aging rather than only relieving symptoms, with reported benefits and reported risks on both sides. This dataset contains one review and nothing else.

Barely there at all

CDP-choline. One mention, and it is a sidebar link on a nootropics vendor’s blog rather than an article about the compound.

Alpha-GPC. Three mentions, all about focus rather than aging. No lifespan or aging data.

Berberine. 10 videos, 4 articles. Promoted as the poor man’s metformin, on the strength of rodent work. No human longevity trial.

Tongkat ali and fadogia. 5 videos, taken for testosterone, with the reported effect described as subtle rather than significant.

Ashwagandha. 27 videos, 2 articles, used for sleep. No aging or lifespan data.

Spermidine. 14 articles, 4 videos, no human trial detail.

L-theanine. 2 articles, 10 videos, paired with caffeine rather than with aging in every mention analyzed.

Rhodiola. 1 article, 2 videos, nothing substantive.

Part 7. Three more claims that do not hold

“Your biological age test tells you how old you really are.” Eric Topol’s position: DNA methylation is the one approach with serious validation behind it, and many tests sold to consumers have not been validated. The company selling you the test also has an interest in the answer it gives you.

“You need eight hours of sleep.” Also contested by Topol. The curve bottoms around seven, and the penalty for oversleeping is steeper than for undersleeping.[16] How regular your sleep is predicts mortality better than how long it is.[17]

“NR raises NAD twice as much as NMN.” The controlled trial that compared them head to head found the two roughly equivalent.[28]

The conclusion

Three things come out of the data.

The research and the content are not about the same subject. The compounds with the best lifespan evidence are prescription drugs, and they appear in almost none of the 285 videos. The compounds with the most airtime are not the ones with the most behind them.

The last three years mostly subtracted. Taurine did not replicate. Collagen’s effect vanished in the trials industry did not fund. Fisetin failed in mice. NAD boosters still raise NAD and still have not shown they do anything with it.

And the field cannot measure the thing it sells. Proving a pill makes healthy people live longer would take decades, and no such trial has been found. So biological age tests stand in for it. In the trial here that ran four of those tests side by side, three years of exercise moved none of them, and the tests disagreed with each other about the same people.[2][7]

Sources

1. https://pubmed.ncbi.nlm.nih.gov/30646252/

2. https://www.nature.com/articles/s43587-024-00793-y

3. https://link.springer.com/article/10.1186/s11556-025-00392-9

4. https://pubmed.ncbi.nlm.nih.gov/28698222/

5. https://academic.oup.com/nutritionreviews/article/81/4/416/6671817 / https://pubmed.ncbi.nlm.nih.gov/36644917/

6. https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2024.1424972/full

7. https://www.sciencemediacentre.org/expert-reaction-to-study-looking-at-omega-3-and-biological-ageing-in-humans/

8. https://www.nejm.org/doi/full/10.1056/NEJMoa1811403

9. https://www.nature.com/articles/s41574-021-00593-z

10. https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2023.1132528/full

11. https://www.cambridge.org/core/journals/british-journal-of-nutrition/article/effect-of-vitamin-d-supplementation-for-major-adverse-cardiovascular-events-a-metaanalysis-based-on-randomized-controlled-trials/8F57F0E377CE0A8FE769C6A9A76F6ED6

12. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0082109

13. https://alz-journals.onlinelibrary.wiley.com/doi/10.1002/alz.12767

14. https://ajcn.nutrition.org/article/S0002-9165(23)66342-7/fulltext

15. https://cosmostrial.org/results/

16. https://www.ahajournals.org/doi/10.1161/JAHA.117.005947

17. https://academic.oup.com/sleep/article/47/1/zsad253/7280269

18. https://www.nature.com/articles/s41593-024-01638-y / https://www.nature.com/articles/s41593-025-01897-3

19. https://academic.oup.com/jcem/article/86/10/4727/2849013

20. https://academic.oup.com/eurheartj/article/38/32/2459/3745109

21. https://academic.oup.com/biomedgerontology/article/80/8/glaf138/8213622

22. https://www.thelancet.com/journals/lanhl/article/PIIS2666-7568(23)00258-1/fulltext

23. https://www.nia.nih.gov/research/dab/interventions-testing-program-itp/publications

24. https://pmc.ncbi.nlm.nih.gov/articles/PMC8508957/

25. https://pmc.ncbi.nlm.nih.gov/articles/PMC9133463/

26. https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2025.1585922/full

27. https://academic.oup.com/biomedgerontology/article/78/1/75/6668639

28. https://www.frontiersin.org/journals/aging/articles/10.3389/fragi.2022.852569/full

29. https://www.nature.com/articles/s42255-025-01421-8

30. https://www.nature.com/articles/s41591-024-03096-2

31. https://pubmed.ncbi.nlm.nih.gov/37289866/

32. https://www.science.org/doi/10.1126/science.adl2116

33. https://irp.nih.gov/news-and-events/in-the-news/irp-researchers-conclude-that-taurine-is-unlikely-to-be-a-good-aging

34. https://www.nature.com/articles/d41586-025-01747-z

35. https://cen.acs.org/analytical-chemistry/biomarkers/Taurine-biomarker-aging/103/web/2025/06

36. https://sciencemediacentre.es/en/taurine-not-reliable-marker-ageing-study-shows

37. https://www.amjmed.com/article/S0002-9343(25)00283-9/abstract

38. https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2025.1729164/full

Video sources

Disclaimer

This is a summary of what other people published. It is not medical advice, and I am not telling anyone to take or stop anything. Some of what is listed here is prescription-only. Talk to your doctor. I do not sell any of this and have no stake in any of it.

So I read back through everything I wrote since last August and looked for what held up. What I still believe after saying it in public and getting hit for it.

Nobody predicts what does not exist yet

Two hundred years ago the serious prediction was mass starvation. The math held up. Fertilizer, crop genetics and modern logistics were not invented yet, so they could not be in the calculation.

In the fifties nuclear power was going to make electricity too cheap to meter. In the late nineties respected economists were saying the internet would matter about as much as the fax machine.

Nobody was stupid. They took the present and drew a straight line forward.

The AI predictions have the same shape. End of work by 2030. End of humanity by 2050. The confidence is the tell.

China is the case I keep thinking about. They believed the population forecast enough to build law on it, ran the one-child policy for decades, and it worked. Fertility is now far below the 2.1 needed to hold a population steady. The country that feared too many people is now planning for too few. The prediction did not just fail. Acting on it created the opposite problem.

I stopped trying to be right about the future. I try to stay able to move when it shows up.

The bottleneck moved and most people are optimizing the old one

Execution used to be the wall. You had an idea, then you needed people, time and money. Now one person can write, design, code, research and ship in a day or week. The tools are not perfect, but they are more than enough.

What is left is deciding. What to do, what to ignore, what to kill.

Meanwhile everyone is still buying execution capacity for a problem they no longer have. Faster call centers instead of removing the need to call. AI answering emails instead of removing the bug. Automation gives you leverage when it removes the step. It gives you noise when it decorates it.

AI does not create character, it amplifies it

I wrote about this three times and got the most heat for it.

If a developer did not read the spec before AI, they still do not read it. They just ship the slop faster. If someone owned their mistakes before, they still do. The tool moved the ceiling, not the person.

Most of the loud criticism about AI mistakes comes from people who never had to be responsible for someone else’s work. Every person I ever hired got something wrong and not once, including me of course, then got something else wrong I had not planned for. The difference between good and bad was never the error count. It was who admitted it and adjusted.

None of that means the tools are magic. Ask ChatGPT to generate an image of a left-handed boy writing with his left hand. You get a right hand almost every time, whatever words you use. Almost everyone in the training images is right-handed, so that is what comes back out. A model can pass a bar exam and fail at something a five year old gets right, because it never really understand it. Although I know adults who still mix up left and right, but that is another story :)

Both things are true at the same time, which is exactly why the accountability sits with whoever is holding the tool. That was the whole argument.

Most people argue with the headline

After that piece the replies came fast. One person said they skimmed it, then gave a firm opinion on what it said. Another announced AI has no right to exist. A third called it dangerous without naming one line from it.

Most read the title, some the subtitle, a few the first paragraph, then go argue in the comments.

I am not complaining, that is the environment. It taught me the title has to carry the real claim, because for most people the title is the whole article. And the ones who read to the end are worth more than the ones who reply.

Your data is not the edge

People kept asking me the same thing after Second Brain (brain.hexact.io) launched. If AI absorbs everything, how long before it absorbs your data?

It already does. That was never where the advantage was.

A fixed dataset stops being an edge the second you upload it. Same file, same model, same answer for anyone. But nothing that matters in a business is fixed. Your pipeline moves. Customers change. Numbers update with every transaction. A three-month-old snapshot of your company is history, not intelligence.

The edge is the current state and the way you connect things nobody else connects. All scientific data is public decades and only some people invent with it.

That is why I built Second Brain the way I did. Not to hide anything from AI. To give it the real current picture instead of fragments guessed from a dozen tools that never talk to each other.

Health: I stopped looking for new information

Calories decide weight. Strength training decides how well you live in your seventies, because losing muscle is what takes away independence. Protein, sleep, consistency. That is the whole list.

Understanding it was never hard. Doing it for years while nothing visible happens for months is the hard part.

The rest is noise sold as a secret. Same as the one cure or superfood story. Bad doctors exist. Broken incentives exist. Antibiotics still work, appendicitis still needs surgery, and life expectancy did not double by accident. Both things are true at once.

The questions that do not close

The piece I liked writing most was the one comparing religion and the simulation hypothesis, and it is the one I am least sure about.

Once you see the pattern you cannot unsee it. A creator or ordering intelligence. A rule set with consequences. Something of you continuing after shutdown. Channels for signaling out and sometimes getting something back. The same skeleton in traditions that never met.

I have no conclusion. I have a preference for patterns over specifics and I am fine not resolving it.

What still bothers me

I spend a lot of time saying the panic about work is overdone, because every previous shift ended fine. It did end fine on average. It was still brutal for the specific people standing in it. The weavers did not become machine operators. Their children did. So the real question is not whether work adapts. It is whether it adapts inside one person’s working life, and I do not know the answer to that.

I argued the bottleneck moved from execution to judgment. That is what people said about every layer right before it got automated. Craft, then process, then execution, each one called irreplaceable until it was replaced. I have no strong reason to believe judgment is the exception, beyond the fact that it is the ground I am currently standing on.

On health I wrote that the basics are simple and settled. People were just as sure that dietary fat was the main problem, and that advice ran for about thirty years before it was corrected. What I trust now is anything with a clear mechanism plus decades of outcome data. That filter is better than most. It would still have passed the fat advice in 1985.

What is actually left

If you cannot predict the future, and you cannot control what technology does to your industry, or which prediction some government decides to act on, then most of what people worry about is outside their hands anyway.

What is left is still huge and it is yours. How you spend the hours. Who you spend them with. Whether the work means something to you beyond the money. Whether you keep your body working long enough to enjoy the years you get. Whether the people close to you would say you were present.

That is the only remaining choice for an individual. Not predicting the future. Not preparing for a version of it that nobody can describe yet. Just figuring out how to live a life that means something in the time you have.

Everything else is commentary.

We were both asking the wrong thing. The question assumes you can pick something that stays useful for years. You cannot, not anymore. Everything moves too fast for that now. What matters is not picking the thing that lasts. It is learning what matters now, and becoming fast enough to learn again when it changes. This is not a question you settle once and shelve. It comes back, again and again, for anyone still going. Me included.

These are my thoughts, not the last word

One note before I go on. Everything here is how I see it, not gospel. I have been wrong plenty, and my opinions have shifted more than once. The core holds. Everything around it keeps moving.

There is endless noise right now. The SpaceX IPO, the next AI model, who replaces whom, which job dies first. I have written about most of that already, whether AI ends us, why you should not trust it blindly. Scroll down through the older posts if you want those arguments. I will not repeat them here. The thing I keep circling back to is quieter, and it runs deeper than a teenager picking a major.

Nobody Is Finished Anymore

Technology and medicine are moving together now, and fast. What if that healthy lifespan stretches by half in our lifetime, maybe more. Not extra years parked in a hospital bed. Good years. Working years. Even adding ten, fifteen, twenty healthy working years changes everything. Sounds like science fiction until you look at what is already being tested on people. Drugs are in trials to slow aging itself. Other teams are clearing out the worn cells that make us old. Most of it may fail. Some of it will not. And it is happening now, in real people.

That quietly breaks the plan most people live by. Study in your twenties. Work until sixty-five. Then wind down. The whole thing assumes a short stretch at the end.

Now add forty years to that stretch. A career used to be one arc, maybe two. That is over. Now it comes in waves. They keep rolling in, and you ride each one as it arrives. It gets harder to run half of a hundred-year life on what you picked up by thirty.

Starting over late is not new. People have rebuilt at forty, at fifty, later. I have done it more than once. But it was usually the exception. The ones forced into it, or the stubborn few who jumped on their own. Everyone else picked a lane and stayed in it for life.

Now the exception becomes the rule. The ground shifts under everyone at once, so most of us will reinvent more than once, planned or not. It stops being a story you admire in someone else. It becomes the ordinary shape of a long life.

Here is the tell. Watch an older man ask my daughter what she wants to become. He asks it like a man who already arrived, as if the question is hers alone and his was closed years ago. It is not. He still has turns ahead, and his are the harder ones. She starts on a blank page. He has to rebuild on top of a life he already chose.

So the people who feel finished are wrong in a brand new way. The old risk was boredom at the end. The new one is going useless in the middle, while the life keeps going. Retirement was designed for shorter lives. Nobody updated the design. The lives changed anyway.

This is why my daughter’s question is everyone’s now. What do I learn next. It does not stop at twenty. It does not stop at fifty.

What learning turns into

Keep learning is old advice. Plenty of people already learn for life, so that is not the new part. The new part is direction. We used to learn in a straight line, deeper and deeper into one field that paid off for decades. That line is gone. Now you master something, the demand moves on, and you have to aim somewhere new. Then again. Then again. The flood of information is bigger than ever, and where you point it matters more than how much of it you pile up.

The way you learn changes too. Here is her real worry. If you can ask ChatGPT anything, why learn at all?

You still learn. Just not the old way.

You do not do long multiplication in your head anymore. The calculator handles that. But you still need to know what multiplication is, and you still need to feel when an answer is off. The basics are what let you catch the machine in a lie.

AI works the same way. Knowing how to ask is the smallest part. The skill that counts is judgment. Knowing what is worth knowing. Knowing when a confident answer is wrong. The tools raised the bar. They did not remove it. That is how I saw it with engineers too, in AI Is Not Replacing Developers.

So you end up needing to understand more, not less. Enough to judge what the machine hands you. That is harder than the old kind of learning.

The best time, and the trap

This is the best time in history to learn anything. The cost fell to almost nothing. A tireless tutor sits in your pocket and explains anything, at any level, as many times as you ask. It is also the best time to do more than ever. One person now does what used to take a whole team.

Here is the trap, and it is mine as much as anyone’s. I keep learning. But I burrow deeper into the same bubble. Same sources, same people who already agree with me.

The bubble does not stop me learning. It quietly decides what I think is worth learning. And the shift that matters almost always starts outside the feed I already read.

You cannot study what you do not know exists. Over a hundred-year working life, that is the real danger, and it is a quiet one. Your field can stop being the field that matters while you keep getting better at it, and you are the last to find out.

What we are building next

We keep building Second Brain. Your knowledge in one place, ready for AI to work with. I already wrote why I think your own data becomes the real edge, so I will skip that here. Right now we are adding more import formats, faster imports, cleaner data management. The boring plumbing that makes everything else possible.

The piece I care about most is still taking shape. An optional assistant that lives on top of your Second Brain.

Most agents today try to do everything and end up doing most of it badly. They are general, and general falls apart the moment the work becomes specific. I think the opposite. Not one assistant that tries to replace you, but a set of professions, each one doing a narrow job well, under your control.

The first one I started on is the one I want most myself. It finds what I do not even know I am missing, the things I would never think to look for. It watches the edge of my bubble and surfaces what I would have walked right past.

For random, open-ended work, tools like Codex or Claude already do a good job on top of your Second Brain. Today that is what they are great at. What we are building is for the other side, specific jobs that have to run again and again and quietly take some of that work off your hands.

I do not have the answer

So what did I tell my daughter.

The truth. I do not know what she should become. I do not know what I should become either, and I have a long stretch left to work it out.

But the direction I am sure about. Stop hunting for the thing that lasts. Learn what is useful now, deep enough to judge the machine, and stay ready to drop it and learn the next thing. Go past your own field while you are at it. A long life spent inside a small view is the thing to fear now, not the work.

I do not know where it lands, and that is the wrong question anyway. It is not going to land, it is going to speed up. I only know the old plan stopped working, so I am learning again.

The predictions sound new.

They are not.

We have heard this before

In 1798, Thomas Malthus predicted mass starvation was unavoidable. In the 1950s, experts predicted nuclear energy would make electricity too cheap to meter. In the 1990s, economists predicted the internet would have no greater impact than the fax machine.

The predictors were not stupid.

They just could not see what did not exist yet.

We never predict what does not exist yet

That is the problem with almost every doomsday prediction. It takes what we know today, projects it forward, and assumes nothing new will show up.

Nobody in 2003 predicted that a social network would reshape elections, mental health, journalism, dating, advertising, and how billions of people form opinions.

Social media created enormous value and enormous damage. It connected the world and damaged attention spans at the same time.

But nobody saw it coming because it did not exist yet.

The future is not a straight line from today.

It never has been.

When doomsday predictions become policy

China took the population crisis prediction seriously. Worried about running out of food and resources, the government introduced the one-child policy and enforced it for decades.

It worked.

Now the fertility rate is far below the level needed to maintain the population. Birth rates collapsed. The country that feared it would have too many people now faces a future where it may not have enough.

That is what happens when you plan the future assuming nothing will change. The predicted crisis does not arrive, but the response creates a different one.

And we still make the same mistake today.

We project current life expectancy, current retirement age, current birth rates, and current health conditions forward and call the result inevitable.

But what if the variables change again?

Life expectancy already more than doubled once. Medical progress is accelerating, not slowing down.

The real question is not whether people live longer. It is whether they stay healthy and productive longer.

If medicine extends functional working age by 20 years, everything changes. Retirement, workforce, healthcare, all of it looks completely different.

We are still projecting the future assuming a 70-year-old in 2050 will resemble a 70-year-old today.

That may become the modern version of the Malthus mistake.

Gene therapies, AI-assisted diagnostics, cancer detection years before symptoms appear, drugs targeting biological aging itself, these are already being researched and tested now.

Maybe they fail.

Maybe they work far better than expected.

As always, we do not know what we do not know.

The black swan is never the one you are watching

Humans are very good at predicting the disaster they already recognize.

Then something else arrives.

We built pandemic playbooks after SARS. COVID still caught the world off guard.

I suspect the same thing will happen with AI. We will spend years arguing about one specific apocalypse scenario, while the real transformation comes from somewhere else nobody modeled properly.

That has happened repeatedly throughout history.

Who knows. Maybe while we are busy debating whether AI will destroy us, aliens show up and AI turns out to be the only thing that saves us. Unlikely? Sure. But no less unlikely than most of the things that actually happened.

Or maybe while we are busy regulating AI, a supervolcano erupts or an asteroid hits and none of our debates mattered at all. The dinosaurs did not have carbon emissions or nuclear weapons. They disappeared anyway. Nature does not wait for our permission. But we are still here. At least that time, we made it.

Most major technologies expanded human capability

They also created new risks. That is how technology works.

The printing press spread knowledge and propaganda. Electricity powers hospitals and electric chairs. The internet gave us infinite information and infinite distraction. Nuclear energy powers cities and can destroy them.

But the direction has still been forward.

People lived to about 32 in 1900. Now it is over 73. Not because of one breakthrough, but because of thousands of small ones building on each other. Clean water, vaccines, antibiotics, surgery, sanitation, nutrition.

Nobody looks back at the industrial revolution and says it was the end of humanity.

It was chaotic. Entire industries disappeared. Millions of jobs changed.

But humans adapted.

That pattern matters.

The doomsday argument has a blind spot

The argument usually goes like this: AI becomes more intelligent, more capable, more autonomous, and eventually replaces humans.

Maybe.

But those predictions almost always assume something unrealistic:

That AI evolves while humans remain unchanged.

That has never happened with any major technology.

Humans absorb tools. We adapt around them. We merge with them.

A person using AI is already more productive than a person without it. That gap will grow.

We spend endless time discussing what AI might become in 10 or 20 years. We almost never discuss what humans become when combined with these systems.

The human of 2040 will not be the human of 2024.

People focus on the creation and forget the creator is changing too.

“AI will take all jobs” is an old panic wearing new clothes

Yes, AI will replace some jobs.

Every major technology shift did.

Typing pools disappeared. Travel agents shrank. Elevator operators vanished. Web developers appeared. Digital marketing appeared.

Entire categories of work were destroyed while new ones emerged.

Work changes shape.

That process is painful for some people every single time. But history does not show work disappearing. It shows people finding new work.

And honestly, humans already outsource thinking constantly. To television. To politics. To religion. To social media feeds. To whatever group they belong to.

AI did not invent intellectual laziness.

We already have everything needed to destroy ourselves

This part of the AI discussion is strange to me.

Humanity has had civilization-ending tools for decades.

Nuclear weapons exist. Biological weapons exist. Chemical weapons exist.

The world could become catastrophic tomorrow without AI being involved at all.

The question was never just about the tool.

It was always about the humans holding it.

And despite everything humans are capable of, we are still here.

We are capable of genocide, torture, mass destruction, and cruelty on unimaginable scales. History is full of evidence.

But history is also full of builders.

Doctors. Engineers. Scientists. Farmers. Parents. Teachers. People quietly improving the world generation after generation.

That side has won more often than not. Not perfectly. Not everywhere. But enough for civilization to keep moving forward.

That track record matters when discussing the future.

And personally, I would rather live in a world where my biggest concern is unexpected AI behavior than a world where a bacterial infection kills people because nobody discovered hand washing yet.

I would rather debate AI replacing jobs than live in a world where childbirth routinely kills mothers and children.

Every era has danger.

Ours is not uniquely terrifying.

It is just modern.

I would rather fight the Terminator than go back to the cave and fight the bear.

We do not even understand ourselves

Here is the part that makes confident AI predictions look strange.

The human brain contains roughly 86 billion neurons connected through an estimated 100 trillion synapses. We still do not fully understand how consciousness works, how memory truly forms, or how subjective experience emerges.

A neuroscientist famously bet in 1998 that science would explain consciousness within 25 years.

He lost the bet in 2023.

The most complex known structure in the universe sits between our ears, and we are still trying to understand it.

So when someone confidently says AI will surpass human intelligence by 2050, the obvious question becomes:

Surpass what exactly?

We do not even fully understand the thing being compared against.

And beyond intelligence, humans have something else entirely. Survival instincts. Intuition. Emotional perception. The ability to function under extreme stress in ways that often surprise even medical professionals.

Anyone who worked in emergency medicine, disaster response, or war zones knows you cannot fit a human into a spreadsheet.

AI learned from what we shared, not from everything that exists

AI models learn from human-produced data.

Text we wrote. Images we uploaded. Code we published. Conversations we recorded.

That is still only a fragment of reality.

It does not include decades of living inside a physical body. Pain. Hunger. Aging. Hormones. Fear. Presence. The way an experienced doctor reads a room before looking at test results. The way a parent senses something is wrong before words are spoken.

AI is extremely good at detecting patterns inside the information it has access to.

But we keep comparing that fragment to the full range of human existence as if they are equivalent things.

They are not.

What happens next

Maybe AI reaches AGI. Maybe it becomes autonomous. Maybe it self-improves faster than humans can react.

Maybe.

But there is another side to that equation.

If humans are capable of creating systems that powerful, imagine what those same breakthroughs do for humans themselves.

The same advances that improve AI will likely improve medicine, science, engineering, education, biology, and human capability too.

Again, people focus on the creation while ignoring that the creator is evolving at the same time.

If you want a creation story parallel: God created people, and look how that turned out. The creator does not always control the creation. Or take the other view, evolution. No creator at all. Just a process that moves forward without anyone steering it. Mutations, adaptation, selection. No plan, no permission, no committee deciding what comes next. It just goes. And here we are, the most complex result of that process, now building the next layer on top of it.

Whether you believe in design or evolution, the conclusion is the same. What comes next is not fully in our control. It never was.

We panic every single time

People protested machines during the industrial revolution and literally destroyed equipment because they feared losing work.

Teachers warned calculators would make students stupid.

My math teacher told me 35 years ago that using calculators would destroy our ability to think.

Then came computers. Then the internet. Then smartphones. Now AI.

The kids who used those calculators grew up to build the internet. The kids who grew up on computers created AI. The thing some people now call superior to humans. So did we become stupid, or did we become capable of building something smarter than anything before us?

The pattern is not that humans stay calm.

The pattern is that humans panic, adapt, absorb the tool, and eventually reorganize society around it.

Every single time.

Security matters more than bureaucracy

I agree with people who say AI safety matters.

But safety and regulation are not the same thing.

Regulation often slows innovation, protects incumbents, creates bureaucracy, and punishes smaller builders more than large corporations.

Security is different.

Security means resilient systems. Better safeguards. Better architecture. Better fail-safe mechanisms.

You do not make cars safer by banning them. You make them safer by building better brakes, seatbelts, and roads.

That approach matters much more long term.

The bigger mistake may be underestimating humans

I am not here to tell you everything will be fine by 2050.

Nobody knows that.

But the conversation around AI is still too narrow. Most discussions isolate AI from the humans using it, adapting to it, and evolving alongside it.

History suggests that is a major mistake.

Every generation believed its danger was uniquely catastrophic.

So far, humans survived plague, famine, industrialization, world wars, nuclear weapons, and the internet.

Not because the risks were fake.

Because humans adapted faster than the predictions expected.

The best way to move forward is not to predict or panic. It is to do something useful. Build something. Solve a real problem. Help someone. The people who actually shape the future are never the ones arguing about it online. They are the ones too busy making things.

And let's be honest. Countries have tried to stop progress before. Ban the technology. Close the borders. Control what people can build. It worked for a while, in isolated places. It never worked at scale. And those places are never the ones anyone wants to live in.

Some people agreed. Some got uncomfortable. And a few revealed the real problem without realizing it.

Most of them did not even read the article.

I do not understand everything either. I never claimed to. That is exactly why I open these conversations and share my opinions, not untouchable truths. But there is a difference between having an opinion and making confident statements about something you have not even read.

People Love to Judge What They Do Not Understand

The responses followed a pattern I have seen for years. Confident opinions from people who did not do the reading.

One person skimmed the article, admitted they skimmed it, and then wrote a firm opinion about what it said. Another declared that AI “has no right to exist.” A third called the piece “dangerous” without addressing a single specific point.

Most people read the title. Some read the subtitle. A few made it through the first paragraph. Then they scrolled to the comments and started arguing. And some just needed to drop a “smart” comment to feed the LinkedIn algorithm. The content does not matter. The engagement does.

The article was clearly about accountability staying with the person using the tool. It also included a section about building a second brain to give AI real context instead of letting it guess. You could call that section a second brain ad if you want. Fair enough. Most people never got to it. They reacted to the headline and filled in the rest with whatever they already believed.

This is how most public discussions work. People do not engage with the argument. They react to the feeling the argument gives them. Then they construct a rebuttal for something that was never said.

The article said accountability stays with the person. Multiple people responded as if it said the opposite. They were not arguing with me. They were arguing with a version of the article they invented in their head.

This happens constantly. People form strong opinions based on partial information, then defend those opinions as if they did the work. They do not know what they do not know. But they are certain about it.

Nobody Was Holding Humans Accountable Either

The idea that human work existed inside some clean accountability system is fiction. Of course, in some cases people do get held accountable. But that is not what this is about.

I have been in business for so many years. I have managed teams across development, operations, sales, production, distribution and finance. Accountability was always the weakest link. Not because people are bad. Because accountability is hard.

A developer ships a bug to production. Who is accountable? The developer who wrote it? The reviewer who missed it? The manager who skipped QA? The PM who compressed the timeline?

In most organizations, the answer is nobody. The bug gets fixed. A retro happens. Everyone nods. The same thing happens again three months later.

So when someone says “humans can be held accountable,” what they actually mean is humans can be blamed. Blame and accountability are not the same thing. Blame finds a target. Accountability changes a system.

The person who is actually accountable is always the same. The owner. The manager. The founder. The one who hired, delegated, and signed off. That was true before AI. It is true now.

AI Made Accountability Visible

The accountability gap that always existed is now impossible to ignore.

When a human employee made a mistake, it was buried inside meetings, email threads, and internal processes. Nobody posted it on LinkedIn. Nobody took a screenshot. It was normalized as “part of working with people.”

When AI makes a mistake, it is public. It is screenshotted. It is shared with commentary about how dangerous this all is. The same mistake that would have been quietly fixed inside a team now becomes evidence in a cultural argument.

AI did not create an accountability problem. It exposed the one that was already there.

Certainty Without Experience

The loudest critics of AI accountability were not operations leaders. Not founders. Not people who manage systems and deal with output quality every day.

When you have done that for years, mistakes do not shock you. They feel familiar. You already know how to review, correct, and iterate. You already know that no resource, human or otherwise, delivers perfect output without oversight.

The people most certain that AI accountability is unsolvable are often the ones with the least experience managing accountability in any form. They judge the situation from the outside, apply logic that sounds right, and skip the part where you actually test it against reality.

That is the pattern. Strong opinions, zero reps.

Real Accountability Was Always on the Manager

When an employee makes a mistake, it is the manager’s responsibility. When AI makes a mistake, it is the user’s responsibility. The tool changed. The ownership did not.

Anyone who has run a team knows this. You hire someone. You give them context. You review their work. If you skip the review and something breaks, that is on you.

AI works the same way. You prompt it. You review the output. You ship it or you do not. If you ship garbage, that is on you, not the tool.

The people who understand delegation are thriving with AI. The people who never learned to manage output are struggling. That gap existed before AI. AI just made it obvious.

“Pro-AI” Is Not a Position

Several people called the article “pro-AI.” As if supporting a technology is an ideology.

I am not pro-AI. I am pro progress. And I do not have any problems with Amish people. If you like that lifestyle, go to the bushes. Being against AI is like being against electricity because sometimes the power goes out or can shock and kill. You can hold that position, it’s up to you.

“You Used AI to Write This”

Of course I use AI to polish my text. That is what tools are for.

Every person who left a comment criticizing AI did it on a computer, using the internet, on a platform built by algorithms. If using AI disqualifies the argument, handwrite your comment and send it by pigeon.

The Acceleration Problem Is Real

AI accelerates everything, including bad decisions. That is true.

If your process is broken, AI will break it faster. If your QA is weak, AI will ship more bugs faster. If your judgment is poor, AI will execute poor judgment at scale.

This is not an argument against AI. It is an argument for better management.

Speed without direction has always been dangerous. A fast car with a bad driver is more dangerous than a slow one. But I personally do not plan to go back to horses.

The answer is better drivers. Better process. Better review. Not slower tools.

Yes, AI Is Dangerous

I am not going to pretend it is not. AI is powerful and power is always dangerous. It can be misused. It can cause damage at scale. It can make bad situations worse faster than anything before it.

But so could every major technology that came before. Nuclear energy. The internet. Cars. Firearms. Medicine. All of them dangerous. All of them still here. Because the value was greater than the risk, and we learned to manage them.

The real question is whether we can handle it. And if we cannot, then maybe we do not deserve to. Every generation faced a technology that could end things if used wrong. That is not new. What is new is the speed. But the answer was never to stop. It was to grow up faster.

What Actually Needs to Change

Nobody knows what AI exactly is. Nobody knows what it will become. Not the people building it, not the people using it, and definitely not the people arguing about it in LinkedIn comments. We are all figuring this out in real time.

The question is not “can AI be held accountable?” No tool can be held accountable. A spreadsheet cannot be held accountable for wrong numbers. A CRM cannot be held accountable for lost deals.

The question is: who owns the output?

If the answer is clear, AI is just another resource. If the answer is unclear, you had a problem before AI showed up.

That confidence is not expertise, it is comfort. And comfort has never been a good guide for understanding anything new.

The conversation about AI mistakes, AI slop, and AI hallucinations is mostly driven by people who have never been responsible for someone else’s output. And I’m not saying AI doesn’t bullshit, of course it does. But they have never hired a person, trained them, watched them fail, corrected them, and then watched them fail again in a completely different way. Why did we never hear this much about human mistakes? Or developers accidentally deleting production databases, crashing servers, and countless other disasters, including not showing up on release day?

If they had managed people, none of this would surprise them.

Every new hire is a challenge, except for edge cases :)

A new employee shows up. You explain the process. They nod. Then they do it wrong. You explain again. They do it differently wrong. Then they make a completely new mistake you did not anticipate.

This does not depend on the role. I have seen it with developers, senior finance managers, operations leads, salespeople, and designers. Everyone makes mistakes. New hires make more. Experienced people make fewer, but sometimes much bigger ones, especially when you stop checking what they do.

The difference is not whether people make mistakes. Everyone does, including me. Some own their errors, learn, and adjust. Others explain, deflect, and repeat. That is the real difference.

I am not blaming anyone. It is just reality. Human work is imperfect.

Now replace “new employee” with “AI agent” and suddenly people are shocked.

The people screaming about AI slop

The feed is full of screenshots of AI getting things wrong: a security hole, a wrong date, a summary that missed the point.

People post them and add, “this is what people want to replace us with.” Or they frame vibe coded product as a liability. Really? So when businesses were completely dependent on developers, that was not a liability?

That is not quality control. That is fear dressed up as criticism. They are not posting because they want AI to get better. They are posting because they want to prove AI cannot replace them.

Meanwhile, real engineers are simply taking advantage of the tools and increasing their output. They are not wasting time trying to prove AI is imperfect. They already know humans are imperfect too.

I have worked with developers my entire career. I never learned to code. I was always the one asking why things were not shipped yet. I have seen human developers make mistakes far worse than anything AI produces today. Specs half-read. Edge cases ignored. Bugs shipped to production because nobody tested properly. Features built wrong because the developer misunderstood the requirement.

Nobody posted screenshots of those mistakes on LinkedIn. Human mistakes were normalized. AI mistakes are amplified.

You are responsible either way

When an employee makes a mistake, it is your responsibility. You hired them. You trained them. You gave them the task. You reviewed the output, or you did not.

AI works the same way. You gave it the prompt. You reviewed the output, or you did not. You shipped it, or you caught it.

The tool is not the problem. Ownership is.

The computer will never replace paper

Thirty years ago, people resisted computers in business. Paper was considered safer. You could see it and touch it. If the power went out, your records were still there.

And they were right about the risks. Computers crash. Power goes out. But nobody talked about the mistakes made on paper systems either.

And yet, nobody runs serious businesses on paper anymore. The risks stayed. The value just became too large to ignore.

We are in the exact same moment with AI.

AI is not your replacement. It is your new hire.

People compare AI to a finished product. They expect zero mistakes. They expect it to understand context, nuance, and intention on the first try.

No employee does that. Not on day one. Some never do.

You give AI instructions. You check its work. You correct it. Managers who know how to delegate and review are getting enormous value from AI right now. The people struggling most with AI are often the same people who struggled managing humans.

The skill that matters is not prompting. It is management.

AI in the hands of entrepreneurs

The biggest value from AI right now is going to entrepreneurs who already know how to manage imperfect resources.

I built a full CRM for my service business in one weekend using AI. It took several more weekends to refine and stabilize. We have been using it for almost a year, and believe me, it is not a liability at all.

Before AI, I would not even have attempted this. The best ready-made solution we evaluated would have taken three weeks to implement.

Entrepreneurs are used to things breaking. Used to adjusting and keeping things moving. Managing an AI that makes mistakes is not scary. It is familiar. And much faster.

The people who struggle most with AI are the ones who have always been executors, never managers. They compare AI output to their own work and feel threatened.

Why a Second Brain changes the equation

AI without context is a smart stranger. It can write, analyze, and generate, but it does not know your business, your customers, your decisions, or your history. That is why it hallucinates. It fills gaps with guesses because it has no real information to work with.

The fix for employees was always the same: give them access to the company’s knowledge, customer data, past decisions, training. The better the onboarding, the fewer the mistakes.

AI works the same way. When you connect it to your own knowledge base, your own data, and your own history, the output changes dramatically. It stops guessing and starts working with real information.

This is why we built a second brain. Not as another note-taking tool with AI or a wrapper, but as the foundation that makes AI tools more effective.

My second brain holds customers, contacts, articles, meeting notes, ideas, business data, old emails, and files. When I ask AI a question, it looks at my actual data. When I want to write, it reads what I have already written. When I need to make a decision, it pulls up context accumulated over years.

The result is not a perfect employee. It is a well-onboarded one. One that still makes mistakes, but far fewer, because it is working with real information instead of filling gaps with noise. It also consumes less tokens, when AI has the right context, it does not waste cycles guessing and generating output you have to throw away.

AI without a second brain is like hiring someone and never training them. AI with a second brain is like giving them access to everything you know.

That is the difference between a tool that frustrates you and one that makes you more effective and productive.

“If the remaining edge is our own data, how long before AI absorbs that too?”

Fair question. And the short answer is: it already does. Today’s AI tools absorb everything that is possible to absorb and learn on that. Not only your data but everything else available.

So if AI can already consume your data, where is the edge?

The edge was never the data itself. The edge is that your data changes. Your edge is the part that is not available to the public and not the static one. Your new data, your decisions, how you correlate the data, how you ask questions on the same data, how you connect even the existing information in ways nobody else does.

If your competitive advantage is a fixed dataset, a document, a report, a one-time export, then yes, that advantage evaporates the moment you upload it. Anyone with the same model and the same file gets the same output.

But the data that actually matters in a business is never fixed. Your customer list changes daily. Your revenue numbers update with every transaction. Your pipeline moves. Your market moves. A snapshot of your business from three months ago is not intelligence. It is history.

The same way all scientific data is available to everyone, but the people who invent look at it differently. Everyone today can create software, but only few actually ship something that someone is willing to use. The difference is what you know more than others right now, and how you connect what you know.

How AI Actually Handles Your Data

Right now, every major AI tool treats your data as a one-time input. You upload a file. The model reads it. You get a response. Tomorrow, that file is outdated and the model doesn’t know it.

Memory features don’t solve this. They store surface-level preferences, your name, a few facts. The gap between “I remember you run a SaaS company” and “your MRR dropped 6% this month and here are the last email messages of the three accounts that churned” is massive. One is personalization. The other requires access to data that changed yesterday.

Most people focus on whether AI can read their information. The harder question is whether AI can stay current with information that never stops moving. Every AI tool has a training data cutoff date, and you do not control how fresh that data is. Unless you bring your own.

The Privacy Illusion

AI companies say they do not train on your data. Your conversations are private. Your uploads are not shared.

Let’s be honest.

When you send an email, that email travels through servers you do not control. When you join a Zoom call, that audio is processed somewhere. When you use your accounting software, your banking app, your CRM, your data lives on infrastructure owned by someone else. When you message on Slack, those words exist in databases you will never see.

Do you really think none of that data is ever shared, aggregated, or used? Not today maybe, but ever? Companies change policies. Companies get acquired. Companies get breached.

But even if we take every privacy promise at face value, it still does not matter for the edge argument. The model is the same for everyone. The data you bring to it is the only variable. The edge is not about keeping data secret. The edge is about how connected it is and how you use it compared to the person who doesn’t.

Control and Freshness

When you use ChatGPT or Claude through their standard interfaces, someone else decides how much context the model retains, when memory resets, whether your uploaded file persists or disappears. Someone else decides what “memory” means. You are a guest in someone else’s system, a system that everyone else has access to.

When you own the data layer, you decide what goes in, how often it updates, what sources feed it, and how it is structured.

A doctor does not diagnose you based on your blood work from 2019. An assistant that operates on stale data is not an assistant. Fresh data produces better answers than a better model working with old data. The model is interchangeable. Your knowledge base is not.

The Answer

Will AI absorb your data? Yes. Models will get better at processing, retaining, and reasoning over user data.

But absorbing a static file is different from maintaining a live, structured knowledge base that reflects the current state of your business. The person who has a local database connected to their payment processor, their CRM, their project tools, their scraped market data, and lets AI query all of it, that person is operating with a fundamentally different tool than someone who pastes a CSV into AI.

Same model. Completely different output.

The edge is not having data. The edge is keeping it alive.

And yes, still keep the personal stuff and the secrets outside of it ;)

That data is valuable. Not in the abstract, privacy-panic sense. Valuable in the sense that it took you hundreds of hours to produce, and it represents real context that no other system has about you.

Now try getting it back. Share me your experience :)

It is a huge question if you will ever get it back, in which format, and when. And even if you do get it back, it will not be what you expected and not in any usable format.

Why I Started Building Second Brain

I have been running multiple businesses for over 25 years. Different industries, different teams, different data, different software and databases. At some point I realized I was repeating context across tools constantly.

My CRM knows my customers but not my investors. My email knows my conversations but not my business metrics. My project management tool knows my tasks but not my articles. ChatGPT knew fragments of everything because I talked to it about all of it, but it forgot between sessions and I couldn’t search or connect any of it.

I wrote about this earlier when I showed how to build a personal knowledge base using Hexomatic and ChatGPT. That was the first version of the idea: collect your public digital footprint, feed it to AI, and get smarter answers. It worked, but it was still a per-task approach, not organized and not built for the long term.

The turning point was when I tried to connect the dots across my own data. I wanted to ask a simple question: “Which of our paid users subscribed to any of my newsletters, and which of our early investors are engaged in any of our products or business?” That data existed, but in three completely different systems. No tool could answer it without manual work.

So I built the answer myself. A local SQLite database on my machine, connected to Claude via MCP (Model Context Protocol). My contacts, investors, customers, articles, ideas, email threads, meeting notes, all in one place. All queryable. All mine.

The Fragmentation Problem Nobody Talks About

Every platform is adding AI now. Gmail has AI. Notion has AI. Salesforce has AI. Slack has AI. Each one is smart within its own walls.

But none of them talk to each other.

Gmail’s AI knows your emails but has no idea what’s in your CRM or Stripe transactions. Notion’s AI knows your docs but can’t see your last published articles and their engagement metrics. Your project management AI knows your sprints but not your customer feedback and sales.

So what do people do? They connect apps. Zapier. Make. n8n. Another integration layer. Then another. Then a dashboard to monitor the integrations. Then a tool to debug when integrations break.

I got tired of it.

The second brain approach is different. One local database. You control what goes in. The AI model you use (today it is Claude, but I don’t think it will be limited to that in the future) can access it through a standard protocol. No middlemen. No sync issues. No wondering which platform has the latest version of your data. And the most important part: you can still access all your data with our Second Brain tool directly, browse the data tables, run queries, or export everything in minutes.

What Actually Changes

When your knowledge base is local, two things happen.

First, you can ask cross-domain questions. "Show me every customer who responded to my last newsletter and also has an active Stripe subscription of product X." That query touches three data sources. In the fragmented world, it's a project. In a local KB, it's one SQL statement. And in the current case, you just chat and get the answer. No need to know anything about SQL. This is what Second Brain is. It is still in its early stage, but just imagine what it is about.

Second, you stop losing your data. Every conversation, every decision, every half-formed thought, it goes into the KB if you want it to. Not as a chat log buried in an export file. As a structured entry you can search, connect, and build on.

We are now working on more import formats, including Notion, Evernote, Trello, and others, so you can bring your existing data in the easiest way.

Vibe coding: 100,000x faster. Vibe debugging: 1x? 2x?

Shipped in 20 minutes. Still fixing it three weeks later.

I have been in tech for more than 25 years. I never learned programming. I was always the annoying guy asking, “Why is this not shipped yet?”

For decades I worked with developers. Some were excellent. Some were careless. Many were somewhere in the middle. And yes, I often struggled with inattentiveness and low ownership. Specs half read. Edge cases ignored. Debugging treated like a chore.

Now the same crowd screams that vibe coding produces garbage code. That AI cannot debug. That junior developers will destroy everything. That the last 1 percent takes forever.

Let’s be honest.

The last 1 percent always took forever.

The Classic Complaints About Vibe Coding

Here is what people usually say:

• AI writes messy, unstructured code

• It does not understand architecture

• It hallucinates APIs

• It ignores edge cases

• Debugging takes longer than writing

• Tech debt will explode

• Security will suffer

This is true, AI models do hallucinate. AI speeds up output, it does not guarantee correctness.

But here is the part nobody wants to say out loud.

If a developer did not care about the product before AI, giving them AI will not suddenly make them care. It will just amplify whatever was already there.

Before, they at least had to read the first sentence of the specification. Now they can paste the whole thing into a prompt and move on. If they had weak understanding before, now the gap is hidden behind fast output.

That is not an AI problem. That is a human problem.

Debugging Was Always a Nightmare

People talk about vibe debugging like it is a new disaster.

Debugging has always been painful. Anyone who has shipped real software knows this. Some bugs are trivial. Some take hours. Some take days. Some only appear in production under weird conditions that nobody predicted.

Nothing new here.

When you deeply understand your product, debugging becomes directional. You know where to look. You know what part of the logic is sensitive. You can tell the AI, “Check the payment webhook handler,” instead of asking it to magically fix everything.

If you do not understand your product, debugging is chaos, with or without AI.

What Changed For Me

Here is what changed.

I started building software myself.

Not toy projects. Real software. With users. With payments. With data. With consequences.

Last year I built a full CRM for my service business in one weekend. Not 20 minutes. More than 20 hours. I started Friday evening. On Monday we were using it in real operations. We imported our existing spreadsheet data. We processed jobs. We tracked payments.

Was it perfect? No.

It took another three or four weekends to refine it. To fix edge cases. To polish flows. To make it stable. We have been using it for almost a year now.

At the same time, we were evaluating ready solutions. The best offer was three weeks to implement. No data import included. And the cost was not small. I am not even focusing on money here. Just time.

For me it was not 100,000x faster.

It was from zero to one.

Before AI, I would not even attempt this. I would need a developer. Budget. Sprint planning. Delays. Pain. Negotiation.

Now I can move.

From Founder to Builder

For 25 years I translated ideas into tickets. Now I translate ideas directly into working systems.

This does not mean I replaced engineers. Good engineers are still gold. Architecture matters. Security matters. Scalability matters.

But internal tools? Operational systems? MVPs? Automation?

The barrier collapsed.

AI can increase developer productivity by 20 to 45 percent in certain tasks like code generation and documentation. In my case, the impact was binary. It was not 30 percent faster. It was possible versus impossible.

That is the real shift.

Why Some Teams Get Worse With AI

If a team does not care about the product, AI will make it worse.

If developers treat tickets as chores, they will now generate more code with less understanding.

If nobody owns outcomes, only output, you will get faster garbage.

AI amplifies intent.

Strong ownership plus AI equals leverage.

Weak ownership plus AI equals noise.

That is why some people experience vibe coding as magic and others as disaster.

The 99 Percent Problem

The meme about being stuck at 99 percent is funny because it is true.

The last 1 percent is thinking. Not typing.

AI automates typing. It does not automate understanding your business model, your users, your edge cases, your constraints.

In my CRM weekend, the hard part was not generating forms or database tables. The hard part was defining exactly how jobs flow through the system. What happens if a client cancels. How and when statuses change.

That is product thinking.

AI did not replace that. It executed it.

Vibe coding and “Real Code”

People say, “Real systems are not vibe coded.”

Are you sure?

Many serious companies now use AI tools in daily development. The code does not magically become fake because AI helped write it. What matters is review, testing, ownership.

The quality bar has always depended on the people maintaining the system.

The Real Divide

The divide is not between real engineers and vibe coders.

The divide is between:

• People who understand the problem and care about the outcome

• People who just execute tasks

AI gives superpowers to the first group. It exposes the second.

For me, vibe coding is not about speed memes. It is about agency. I can build what my business needs without waiting for someone else to prioritize it.

It is exciting. It is empowering. It is messy sometimes. But it is real.

Nobody Banned Mediocrity

Let me say this very simply.

Nobody banned mediocrity.

AI did not create it. It was always there.

Years ago, many accountants said computers would never replace paper. They said real bookkeeping must stay on paper. That serious work cannot happen on a screen.

The tool changed. The standard did not disappear. The market rewarded those who adapted and left the rest behind.

It is the same now.

Some developers will use AI to produce weak systems faster. Some founders will ship half-baked products and blame the tool. Some teams will pile up messy code.

That is not new.

AI lowers the entry barrier. It does not lower the bar for quality. If anything, the bar is higher now. Because speed is no longer the excuse.

If you understand your product and care about the result, AI is leverage.

If you do not, it will only expose it.

Ride-hailing replaced taxis.

Food delivery replaced calling restaurants.

E-commerce replaced ordering from catalogs.

Online booking replaced emailing hotels and travel agencies.

Most people assume the breakthrough was speed.

It was not.

Cars did not suddenly drive faster.

Kitchens did not start cooking twice as quickly.

Trucks did not break the laws of physics.

What changed was uncertainty.

The real breakthrough was removing the back-and-forth.

You stopped calling.

You stopped negotiating.

You stopped repeating your details.

Instead, you saw:

Who is coming.

Where they are.

When they will arrive.

What happens next.

That visibility changed behavior more than speed ever could.

Stop Automating the Wrong Layer

Many businesses try to improve outdated processes by automating them.

Smarter call centers.

AI answering calls.

AI replying to emails.

Chatbots handling messages.

But pause for a second.

When was the last time you called a taxi?

When was the last time you called a hotel and dictated your passport details and credit card over the phone?

The companies that won did not build better call centers.

They removed the need for them.

Email did not make fax machines faster. It replaced fax.

E-signatures did not improve printing and scanning. They removed both.

It is like building a tool that helps you copy and paste faster instead of building a system where data syncs automatically and copying disappears.

Automation creates real leverage when it removes the step itself.

Not when it decorates it.

Not Everything Should Disappear

This does not mean calls should vanish everywhere.

There will always be complex situations.

Major renovations.

Custom projects.

Edge cases that require discussion and inspection.

Back-and-forth is not wrong.

In many scenarios, it is necessary.

And if a process must exist, improving it makes sense.

The point is more specific.

Some categories no longer require that layer at all.

Booking a flight does not require emailing the airline.

Ordering food does not require calling the restaurant.

Booking a hotel does not require sending documents manually.

Getting a taxi does not require negotiating with dispatch.

Those steps were not optimized.

They were removed.

Urgent services contain many situations that are closer to ordering a taxi than to planning a custom renovation.

A clogged sink.

A dead battery.

A jammed lock.

A grill that will not ignite.

A toilet that will not flush.

These are defined problems.

They do not require five calls, repeated explanations, or layers of coordination. Most of them are a one-person job, one technician, one visit, one fix.

Complex work will always exist, custom builds, renovations, large installs.

But most urgent fixes are not complex. They are simple one-man jobs trapped inside a coordination model designed for multi-team projects.

Urgent Services Are Still Operating Like This

When something breaks, the pattern is predictable. You search online, submit a form, call a number, leave a voicemail, explain the problem, send photos, wait for a callback, answer the same questions again, discuss timing, ask for a rough estimate, and hope someone actually shows up.

And sometimes the job is so small that they simply decline it. Not worth the trip. Not worth blocking the schedule.

Most of the stress is not the repair itself. It is the uncertainty.

Will they respond today?

Will they fit you in?

Will they increase the price on arrival?

Will they cancel?

Will they even accept the job?

The delay is often tolerable.

The uncertainty is what drains you.

Inaccessibility Shrinks the Market

Every home and office has small problems that stay unfixed for months. A loose hinge, a drawer that sticks, a switch that works only sometimes, a slow leak under the sink, an appliance that runs but not properly. None of these are major projects. They are simple fixes.

They remain unresolved not because they are expensive or technically difficult, but because arranging the repair feels heavier than the problem itself. You think about searching, calling, waiting, explaining, negotiating a time slot. For something small, it feels disproportionate.

So you postpone. Or you replace.

You buy a new chair instead of tightening the frame. You replace a faucet instead of swapping a cartridge. You discard an appliance that needed a minor adjustment. The friction of access pushes people toward replacement instead of repair.

That creates waste. It increases long-term cost. And it quietly shrinks the repair economy.

When access becomes simple and predictable, behavior changes. People fix small things earlier. More jobs get done. Accessibility does not just improve convenience. It expands the market itself.

Stress Is Not a Minor Detail

When something breaks at the wrong moment, it affects more than the object.

The sink leaks late at night.

The toilet clogs during a family gathering.

The front door lock jams when everyone is in a rush.

Guests are arriving and the grill does not start.

Tension rises fast.

Plans shift.

People argue.

Blame appears.

The mechanical issue may be small.

The emotional impact is not.

Uncertainty multiplies stress inside families and businesses.

What Urgify Is Actually Building

Urgify is not about making technicians physically faster.

It is not about AI answering more calls.

It is about removing the unnecessary coordination layer.

No browsing endless profiles.

No calling five numbers.

No negotiating prices in the moment.

No vague promises.

Instead:

A request is created.

A provider accepts it.

The acceptance is visible.

The status is tracked.

Responsibility is clear.

Not automating chaos.

Removing it.

The goal is not more communication.

It is less friction.

Ride-hailing did this.

Food delivery did this.

E-commerce did this.

Urgent local services are simply late to the same structural shift.

The real breakthrough was never speed.

It was certainty.

Certainty comes from removing the step that creates doubt in the first place.

That is why we built Urgify.

Not to create smarter phone trees.

Not to make messaging faster.

Not to optimize negotiation.

But to make that negotiation unnecessary for most defined urgent tasks.

Is it difficult?

Of course.

You are dealing with the physical world.

With real people.

With unpredictable variables.

But is it doable?

One hundred percent yes.

Other industries have already shown the pattern.

Eliminate the unnecessary step.

Make commitment visible.

Use automation to remove friction, not decorate it.

Urgify exists to bring that shift to urgent local services.

And once that shift happens, behavior changes.

We compress uncertainty into stories. We simplify causality. We remove randomness. We rewrite events so they feel coherent, intentional, and explainable.

This is not moral failure. It is a coping mechanism. Reality is messy, probabilistic, and uncomfortable. Stories make it survivable. They make it teachable.

This distortion shows up everywhere. In education. In business. In social media. And now, in how we train machines.

Once you see the pattern, the rest becomes obvious.

1. The Professor Problem

Business education is a clean version of a dirty process.

Students are taught through success stories, frameworks, and confident explanations. Failures are minimized. The uncomfortable parts are reframed as “bad luck” or removed entirely.

The result is predictable. Graduates who can repeat models but struggle to reason under uncertainty.

Real critical thinking is built on what is missing. Constraints. Tradeoffs. Failed experiments. Decisions that were reasonable at the time and still led to collapse. The reasons a good idea died anyway.

Research on entrepreneurial learning consistently shows that failure can be a powerful input, but only when it is confronted directly. Learning depends on reflection, context, and honest processing. When failure is softened to protect ego, it stops teaching.

When educators hide their own failures or rewrite them as abstract “lessons learned,” students lose the most important dataset: what actually happened and why.

The damage is not just incomplete information. It is causal distortion.

Students over-attribute outcomes to intelligence, grit, or vision. They under-attribute to constraints, execution detail, and chance.

If the dataset is curated, the student becomes confidently wrong.

2. The More Dangerous Problem: We Often Don’t Know Why Things Worked

Hiding failure is not the worst issue.

Not knowing why something succeeded is worse.

A startup raises capital. The founder explains it as product-market fit, timing, or traction. The real driver may have been invisible. A fund needed to deploy capital before quarter end. Other deals collapsed. Internal incentives forced a yes.

The founder never sees this. They only see the outcome. So they construct a story that feels true.

This happens everywhere. A hire works out. Was it the process or luck. A launch succeeds. Was it positioning or timing. A deal fails. Was it execution or an unrelated external event.

Humans reconstruct narratives after outcomes. Uncertainty becomes inevitability. This is not dishonesty. It is cognition.

The business world is full of single-sample experiments presented as universal lessons. Someone succeeds once, then teaches a framework. But running an experiment once does not allow you to separate signal from noise.

Maybe the success came from geography. Or timing. Or an introduction they did not earn. Or a competitor imploding quietly.

They will never know. But they will explain it anyway.

3. Social Media as a Bias Engine

Social media is not fake. It is selectively coherent.

It rewards confidence, certainty, and clean narratives. Ambiguity does not travel well.

Nobody posts: “I don’t know why this worked. It might fail tomorrow.”

They post: “Here are the five lessons learned.”

Even honest attempts at reflection require compression. Random events become strategy. Chaos becomes intent.

So the internet is not a record of reality. It is reality filtered for status, incentives, and narrative clarity.

This matters because modern AI systems learn from large volumes of human-generated text. When the text is biased toward polished stories, the model learns polished stories.

Just like the student.

AI is trained on the rewritten version, not the confused, partial, contradictory reality underneath.

4. AI as a Student of Our Edited Selves

The simplest mental model of AI is this.

It is a student that never lived a life but read everything.

The problem is that much of what it reads is performance.

AI learns what humans say happened, not what actually happened. Humans lie constantly. Not always deliberately. Often socially. Often unconsciously.

In machine learning, this shows up as bias in data and bias in models. Training data reflects omissions, selection effects, and social incentives. The model absorbs them.

Better documentation helps. Transparency helps. But none of that fixes the underlying issue.

The source material is already distorted.

5. Why “AGI” Might Never Understand Humans

Many arguments focus on consciousness or embodiment.

A simpler explanation exists.

We are training machines on a version of humanity optimized for approval.

We hide incompetence. Contradictory motivations. Hypocrisy. Confusion about causality. Random events we later call vision. Decisions made for one reason and justified with another.

Most importantly, we hide the moments when we genuinely do not know whether success came from skill or luck.

Lies can sometimes be detected. Confusion cannot.

When humans themselves cannot separate signal from noise in their own lives, there is nothing clean to teach.

We publish press releases. Machines become excellent at writing press releases.

They learn what intelligence sounds like, not what decision-making looks like under pressure.

As long as this continues, AI will misunderstand the most human parts of us. The irrational. The social. The ego-driven. The inconsistent. The private.

Not because it cannot model them. Because we refuse to describe them honestly.

6. The Asymmetry That Remains

Humans retain an advantage over machines, but not for the reasons people usually claim.

Not intelligence. Not creativity. Not consciousness.

The advantage is structural.

Humans do not operate on clean data. We act on incomplete, contradictory, and often wrong explanations of our own behavior. We make decisions without understanding the full causal chain. We revise our reasons after the fact. We pursue outcomes for one reason and justify them with another.

AI learns from the explanations. Humans act inside the confusion.

Machines need consistent signals to optimize. Humans thrive in inconsistency. We change goals mid-stream. We respond to social pressure, fear, ego, and status in ways we do not fully articulate, and often do not even notice.

AI learns the narrative layer. Humans operate below it.

This creates a permanent mismatch.

As long as humans continue to curate truth, simplify causality, and rewrite uncertainty into stories, machines will be trained on a made-up world rather than the one humans actually inhabit.

That world does not exist.

So AI will continue to be highly competent, and consistently wrong.

Not because it lacks capability, but because it is learning from performance instead of practice.

This is why humans will always outmaneuver machines.

Not by being more rational, but by being less legible.

Not as a tactic, but as a condition of being human.

AI is no different.

Right now, a lot of discussion revolves around what can realistically be automated. What tasks are safe. What jobs are next. What AI still cannot do.

That discussion sounds practical. But it quietly misses where the real constraint already is.

Execution Is No Longer the Hard Part

AI removed execution limits faster than humans can adapt.

One person can now do what used to require a full team. Write, design, code, analyze, research, iterate. The tools are not perfect, but they are more than capable.

The bottleneck is no longer execution.

The bottleneck is attention, judgment, and prioritization.

We already see this inside teams and large organizations. Not moving forward is rarely caused by lack of tools or talent. It happens because decision-making cannot keep up with the volume of possibilities. Too many options. Too many paths. Too many things that can be done right now.

Brains choke long before software does.

Process automation is largely solved. The next real shift is decision-making. Choosing what matters. Choosing what to ignore. Choosing what to ship and what to kill.

This pattern repeats across industries.

Tesla Criticism Misses the Category Shift

Look at how people talk about Teslas.

Panel gaps. Plastic interiors. Water leaks. Usually compared to German cars and traditional craftsmanship.

Meanwhile, the car drives itself.

That single fact changes the category. Once autonomy enters the picture, interior trim stops being the defining metric. Judging a self-driving car by legacy standards is like comparing horse carriage upholstery after the engine was invented.

The comparison is not wrong. It is irrelevant.

Autonomy changes what the product even is. And today, no traditional manufacturer is close to matching that capability at scale.

History shows this pattern clearly. When a new category appears, people cling to familiar metrics because they feel safe. Familiarity is comforting. It lets people avoid confronting the implications of change.

The Same Thing Is Happening in Software

Now look at the “vibe coding” debate.

Software engineers mocking AI-generated code. Complaints about structure, cleanliness, style, maintainability. Some of it is valid. Much of it is missing context.

After more than 25 years working with in-house teams and vendors of every size, I can say this without hesitation. Humans have shipped unmaintainable garbage for decades. Production systems full of hacks, shortcuts, copy-paste logic, and bugs no one understands.

AI introduces a new kind of weirdness. But it also removes excuses.

When execution becomes cheap, judgment becomes visible. When anyone can generate code, taste and discipline matter more. Sloppiness gets exposed faster. Competition increases. Comfort disappears.

That pressure is healthy.

It forces developers to think more carefully about architecture, constraints, and outcomes. Not because AI is perfect, but because mediocrity is harder to hide.

The Common Failure Pattern

Across AI automation, self-driving cars, and vibe coding, the pattern is the same.

People obsess over surface flaws while the underlying category shifts.

They argue about polish while leverage explodes.

They debate aesthetics while the cost of iteration collapses.

They defend old standards because those standards once defined their advantage.

History is not kind to that mindset.

The winners are rarely the people who perfect yesterday’s metrics. They are the ones who recognize when the scorecard itself has changed.

The Real Question for 2026

The uncomfortable truth is this.

Most people are no longer blocked by execution. They are blocked by decisions.

What to build.

What to ignore.

What to stop doing.

What actually matters.

AI did not eliminate the need for humans. It raised the bar on judgment.

So the real question going into 2026 is not what AI can or cannot do.

It is this.

What is actually bottlenecking you now.

Execution. Or decisions, taste, and judgment.

Someone discovers a new “curing thing”. A pill, a compound, a diet, a peptide, a plant extract. Supposedly it cures almost everything. Cancer, inflammation, depression, aging, sometimes even bad posture and bad luck.

And then comes the familiar explanation.

Doctors are hiding it.

Big pharma is blocking it.

Governments are evil.

Everyone is greedy.

Of course, there are bad people. There are bad doctors, sometimes even criminals. There are corporations that chase profit at the expense of ethics. Anyone who says otherwise is naive. But stopping the conversation there is lazy thinking. People who accept these simple explanations too easily are often influenced by a well-crafted marketing narrative rather than a real look at reality.

Let’s talk about a few uncomfortable realities.

1. Yes, the healthcare system is flawed. And yes, it still saves lives.

Almost everyone I know has a story. A wrong diagnosis. A prescription that made things worse. A doctor who did not listen. Sometimes the consequences were lethal.

Healthcare systems are bureaucratic. Incentives are often broken. Mistakes happen more than they should.

And yet.

When your child has a high fever and nothing helps, you go to a doctor.

When someone has appendicitis, pills from TikTok will not help.

When a bacterial infection hits, antibiotics work. Period.

Many doctors do an excellent job. Many drugs work exactly as intended. Many companies have created treatments that radically changed human survival and quality of life. And let’s not forget that average human life expectancy was around 30–40 years at the beginning of the 20th century, and today it is over 70 years globally, more than double in just one century.

Both things can be true at the same time. A system can be flawed and still essential.

2. Living systems are not simple machines

This is the part most miracle-cure narratives ignore.

The human body is not a flashlight where you change a battery and everything works again.

Each cell is a factory. Each system is interconnected. Hormones, immune responses, microbiome, genetics, environment, stress, sleep, nutrition. Everything affects everything.

A substance that helps one pathway can harm another.

A dose that helps one person can hurt another.

Short-term improvement can hide long-term damage.

Biology is messy. Nuanced. Probabilistic.

The idea that one tablet can “fix everything” is not optimistic. It is scientifically childish.

If something truly cured everything, it would not be hidden. It would be impossible to hide. And of course there are always new things that are mostly not yet accepted, and sometimes they become something real and useful.

3. Reading an article does not make you an expert

We live in a time where access to information is confused with understanding.

Someone reads three blog posts, watches a podcast, and suddenly knows more than people who spent decades studying a field, especially when it comes to the human body and living organisms.

Science does not work like that.

Most scientific questions do not have final answers. They have current best explanations. These evolve. They get refined. Sometimes they get replaced.

That does not make science weak. It is just what it is, and it is real.

Certainty is usually a red flag. Especially when it comes with a product link.

4. We have seen this cycle before, many times

This is not new.

Years ago, cholesterol was the enemy. Fat was bad. Eggs were dangerous. Butter or meat was almost treated like poison.

Then the narrative shifted. Sugar became the villain, even though hundreds of years ago it was saving lives from hunger. Then carbs. Then gluten. Suddenly gluten was the root of everything, even for people without celiac disease.

Later it was seed oils. Then lectins. Then oxalates. Then histamines. Now it is peptides, compounds, injections, protocols.

Each wave follows the same pattern.

A complex problem is reduced to one cause.

A single fix is promoted as universal.

Dissent is framed as corruption or ignorance.

Sometimes these trends contain a grain of truth, blown up by marketing. Often they help a subset of people. Almost never are they universal.

Human biology did not suddenly change every five years. The story did.

5. “Natural” does not mean safe, and “synthetic” does not mean dangerous

One of the most common tricks in miracle-cure narratives is language.

“Natural” is framed as good.

“Chemical” or “synthetic” is framed as bad.

This is marketing, not biology.

Arsenic and sulfides are natural. So is cyanide.

Insulin is manufactured. So are vaccines.

What matters is not where a substance comes from, but how it interacts with the body, at what dose, for how long, and in which context.

Nature does not optimize for human safety. It optimizes for survival.

6. Correlation feels convincing. Causation is hard.

Another reason miracle cures spread so fast is personal stories.

“I took X and I felt better.”

“My inflammation disappeared.”

“My friend reversed everything.”

These experiences feel real because they are real to the person experiencing them.

I always remember my medical university years, when one of my professors used to say, “I believe it happened after taking that, but prove to me that it was because of that.”

Symptoms fluctuate. Placebo effects are powerful and measurable. Lifestyle changes often happen alongside the “cure” and get ignored.

Science tries to separate coincidence from cause. That process is slow and boring.

Stories are fast and emotionally satisfying. That is why they dominate the internet.

7. If something works, it eventually gets adopted

There is a belief that effective treatments are always suppressed forever.

History does not support this.

Treatments that work eventually spread because doctors want better outcomes, researchers want recognition, and companies want profit.

That alignment is imperfect, but it exists.

This is why antibiotics exist.

This is why insulin exists.

This is why anesthesia exists.

Suppression narratives assume a level of global coordination that simply does not exist.

8. The internet rewards certainty, not accuracy

Online platforms reward confidence, not precision.

A video titled “This might help some people under specific conditions” will not go viral.

“This cures everything they don’t want you to know” will.

Algorithms amplify emotional certainty. Fear, anger, and outrage travel faster than careful explanation.

Over time, nuance loses.

9. Wanting simple answers is human

People are not stupid for wanting simple solutions.

Chronic illness is exhausting. The healthcare system can feel cold and transactional. Many people feel unheard.

A single cure offers comfort. It creates villains and heroes. It restores a sense of control. What rarely gets discussed are the lives lost or the conditions made worse when people fully commit to these “single cure” campaigns and delay or abandon real treatment.

That emotional need deserves empathy.

But empathy should not turn into blind acceptance.

10. Science is not truth. It is a process.

This mistake is not limited to health.

When AI tools appeared, many people treated their answers as absolute truth. Final. Non-discussable.

But AI does not produce truth. It produces likely answers based on existing and popular data.

Science works the same way. Models evolve. Theories change. Knowledge expands.

Anyone selling certainty is not educating you. They are selling relief from uncertainty.

And uncertainty is part of reality.

Food delivery went from calling a random restaurant to tapping two buttons.

Transportation went from “Do you have a taxi number?” to a car arriving in minutes.

Hotels and rentals became instant, transparent digital bookings.

But one industry is still stuck in 1998:

Urgent real-world services.

When something breaks: a lock, a pipe, a tire, a battery, a grill, your Wi-Fi - you still follow the same outdated routine:

• Google “handyman near me”

• Call five numbers

• Three don’t answer

• One says “maybe tomorrow”

• One wants pictures, videos, and a full explanation

• And maybe, if you’re lucky, someone eventually shows up one day

And the strangest part?

We all accept this as normal.

The real world is modern everywhere except where it matters most

You can order a sandwich, a car, or a place to stay in seconds.

But fixing a leaking pipe still requires negotiating, guessing prices, and hoping someone is available.

Platforms evolved in food, transportation, accommodations…

but urgent help never got a true upgrade.

Not because people don’t need it, but because the existing tools weren’t built for sudden, unplanned problems.

Marketplaces want you to browse.

Directories want you to call.

Business software wants you to manage a company.

None of them help when you’re standing outside a locked door or staring at water dripping from your ceiling.

Urgent moments require speed, not browsing.

Clarity, not negotiation.

Action, not back-and-forth messaging.

So we built something new.

Introducing Urgify.app - Soft Launch

Urgify connects you with service providers the same way other aggregators do.

Not a marketplace.

Not a directory.

Not B2B software.

Just fast, reliable help when something breaks.

How it works:

You describe the issue →

You select the pricing option →

The nearest available provider is dispatched.

No browsing.

No quotes.

No calling around.

No “maybe next week.”

And for people who want to work for themselves, Urgify delivers real jobs instantly - no bidding, no chasing leads, no unpaid conversations.

This model removes friction for both sides:

customers get help fast, and providers get real work without noise.

Why start small?

Because the goal isn’t hype.

It’s correctness.

Urgify isn’t a marketplace with thousands of random gigs.

It’s a structured, reliable system for urgent situations - so every detail matters.

We are launching this as a controlled experiment, improving daily, learning from real behavior, and expanding step by step.

The vision is simple:

Modern life shouldn’t break every time something breaks.

Urgent help needs to join the rest of the on-demand world.

Soft launch is live. More updates soon.

We’re just getting started.

I wrote most of it months ago. Then I stopped. I told myself it wasn’t ready. That it needed more editing. That maybe I shouldn’t publish it at all. I opened the draft. Closed it. Rewrote a few parts. Ignored it for weeks. This loop repeated too many times. If you’ve ever tried to finish something personal, you know the feeling.

At some point I realized there is no perfect version. A book doesn’t get better because you stare at it longer. It only becomes real when you publish it. So I finished it as it is, rough edges and all. That decision felt honest.

The book is called Bones: 21 Clichés Running and Ruining Your Business. But it’s not really about these twenty-one clichés. We all have our own sets of phrases we repeat without thinking. Some come from our first jobs. Some from managers we once respected. Some from fear. Some from habit. They quietly shape how we act, decide, and explain things to ourselves.

My list won’t match your list. That’s the whole point. The book is meant to make you notice your own patterns, not copy mine.

Every chapter is short. One cliché, a few angles, a few questions. You can agree with a chapter or disagree with it. That tension is what makes it useful. I didn’t try to hand out rules or claim authority. I wanted something simple that invites reflection, not a lecture disguised as a book.

This is why finishing it feels good. The second one is already in progress. I feel more confident moving forward without overthinking every page.

Meanwhile, most people are just doing their jobs. Managing clients, answering emails, caring for lawns, connecting cables, and dealing with real work. Many have no idea what AGI or “vibe coding” even means. Others are nervous about it without knowing exactly why. The story always sounds the same, cheaper, faster, smarter machines will take over, and we’ll all watch from the sidelines. I think it was something similar when cars came to replace horses, and people worried about jobs being wiped out, but they just evolved.

And in the first place, people say AI will replace developers. That’s not what’s happening.

AI isn’t replacing developers. It’s exposing them.

For years, tech has been full of people who called themselves developers but never really built anything meaningful. They managed processes, wrote shallow code, outsourced complexity, and made everything look harder than it was. That kind of work is disappearing.

AI is separating people who think from people who copy. It removes the gap between an idea and execution. You can’t hide behind process anymore.

The developers who care about logic, systems, and design now have a tool that multiplies their output. The ones who built careers around buzzwords and inflated rates are being replaced, not by AI, but by transparency and simple common sense.

I’ve spent most of my life in tech. I built software companies, led teams, and worked closely with engineers. But there was always a barrier. I never learned to code.

That limitation shaped how I worked. I understood the problem but couldn’t directly create the solution. I had to translate ideas through others. Some got it, others didn’t, and progress always depended on someone else’s speed and priorities.

That’s not the case anymore.

AI removed that barrier. I can sit down, describe what I want to build, and watch it come to life. I can test, adjust, and deploy without waiting for anyone. I still don’t call myself a programmer or software engineer, but I can now build software products myself.

It’s not about replacing developers. It’s about removing friction between ideas and action.

When you can move from concept to prototype in a day or two, your entire approach to building changes. You stop waiting for perfect plans and start experimenting. You stop asking for permission and start creating.

AI didn’t end software development. It made it honest.

Real engineers are thriving because they finally have tools that let them focus on solving problems faster. Pretenders are fading because they were never builders to begin with.

This is the transition we’re in. Not mass replacement, but redefinition. The same way digital cameras didn’t kill photography, they exposed who actually understood light and composition. The ones who only knew how to take pictures aren’t needed anymore, any child can do that.

AI is doing the same for engineering. It rewards people who think clearly, build practically, and adapt fast.

If you’ve been sitting on an idea because you “don’t know how to code,” or don’t have the budget to hire software engineers, that excuse doesn’t work anymore. You can learn the basics. You can use AI tools to build. You can ship something real. And of course, you won’t manage to scale alone, and nobody is planning to.

You don’t need to become an engineer. But you should think like one.

AI is already filtering who does and who doesn’t.

That’s the real change, not job loss, but truth.

The “founder journey.” The “light and dark sides of entrepreneurship.” The “everything costs something” speech.

They sound deep, but most of them are motivational packaging, not reality. Many come from people who want to look like founders more than they want to build something real.

You can spot it easily. Endless posts about fundraising rounds, team culture, resilience, morning routines, founder depression, mentorship quotes, and photos from investor meetings. What’s missing is the part where you actually solve a problem someone pays for.

The 10-Year Myth

One of the most common claims is that entrepreneurship is a journey limited by years. That’s not how it works. If it ends after 10 or 15 years, you either quit, retired, or the company shut down. Nothing wrong with that. But otherwise, the road doesn’t stop.

The stakes rise. The problems evolve. But they never disappear.

Each new level brings its own version of chaos. You don’t finish the game. You just play on harder settings.

The Light Side Illusion

There’s constant talk about the bright side — funding rounds, prizes, PR, “world-class teams,” or showing off company size. I did the same. Most founders do at some point. But notice how rarely anyone shares real numbers — revenue, profit, or cash flow — while preaching about transparency and honesty. And don’t say it’s a “commercial secret.” In most cases, it’s not. It’s just uncomfortable to admit that growth stories often look better on slides than in reality.

That’s because the light side is easier to talk about than the real work.

True founders don’t chase applause. They build because they can’t not build. Their mind never stops. They see broken systems and think about how to fix them. They are driven by something they often can’t explain.

It’s not about headlines or hype. Those moments feel good, but they fade fast. The danger is when younger founders start believing that’s the goal and begin chasing optics instead of outcomes. That’s how many get stuck in the illusion of progress.

The Dark Side Drama

Then there’s the opposite extreme — the dark side. Missed birthdays. No balance. Anxiety. Burnout. It’s often shared like a badge of authenticity.

But not everyone burns out. Some founders like the squeeze. They find energy in pressure. Others can’t handle it. There’s no fixed formula.

In the beginning, most of us don’t realize how important family is. Over time, you learn to balance it the same way you learn to balance a business. You make mistakes, you adjust, you improve what matters.

If it feels like pain for too long, something is broken. Either the business is off track, or you are forcing yourself into the wrong kind of work.

Entrepreneurship is not meant to be suffering. It’s creation under pressure. The difference is how you digest the pressure — as pain or as part of reality. Being an entrepreneur is not about being a victim of suffering. It’s probably the most self-chosen path there is, so if you really suffer, then it’s not yours. And if you don’t, stop acting like a victim. :)

The Fundraising Trap

Funding updates dominate founder conversations. Many treat raising a round like crossing the finish line. But raising money only means someone believed your story. It’s part of the process, not proof of success.

There is nothing bad about funding. In some cases, it’s the only way to move forward. The problem starts when raising money becomes the goal instead of the tool.

It doesn’t necessarily mean the product works or that customers care.

Real validation happens when people pay — and keep paying. That’s the market speaking. Everything else is noise.

Too many founders spend years pitching investors instead of building products. They master decks, not solutions. I went through that phase too. It was easier to raise money than to build something people loved and paid for. I’m glad I learned that lesson early. The real game begins when the customer pulls out a card, not when an investor wires funds.

The “Founder Persona” Problem

We now have a generation performing the founder role. They post like founders, dress like founders, talk about vision and grind, but rarely sell anything. It’s a performance.

Even many real founders fall into the same trap. They start counting down to exit, planning retirement from the “founder life.” Nothing wrong with that, but for a real entrepreneur, the drive never disappears. It’s not a job. It’s a mindset.

Being a founder has become a badge, not a result of solving something valuable. That’s the distortion. You don’t become a founder because you post about resilience. You become one because you make something useful exist.

And about the “changing the world” narrative — there are millions of entrepreneurs building successful businesses without “changing the world.”

The Real Constant

The path doesn’t get easier. It only changes shape.

At $10K in revenue, you worry about paying bills.

At $10M, you worry about people and systems.

At $100M, you worry about markets, politics, and risk.

Different level, same tension.

I think the only right way to move forward is to ask yourself: do you enjoy solving real problems, or are you chasing validation and attention?

If it’s the latter, the dark side always wins.

The Real Filter

Being a founder or entrepreneur isn’t a business plan, a funding milestone, or a motivational checklist. It’s not about showing your best side online or building a personal brand around quotes.

It’s about wanting to keep building when no one’s watching, when the hype is gone, and when the next problem shows up.

There is no finish line. No “one big result.” It’s an infinite process. You go up and down, again and again.

And that’s what makes it real. That’s exactly what makes it exciting.

I’ve learned this the long way. Through failures, pivots, exits, and restarts. The pattern repeats, but the drive never fades. The real reward isn’t the outcome, it’s the ability to keep playing the game and still care enough to build again. And yes, the monetary reward always comes when you create real value — maybe not immediately, but always eventually.

I have been deep in this for many years. My medical background taught me to look past the trends and focus on the core principles. The human body has rules. Looking for shortcuts, miracle programs, or “magic winds” is useless.

From one hand it is very easy to understand these principles, but in real life it is not so easy to implement them consistently. Still, the best approach is to focus only on what is important and works. In that way you will always achieve results, even if progress feels slow.

This is a summary of what I found most important for myself. It is not a prescription or medical advice. It is the structure I use, and I will list the references at the end for anyone who wants proof.

Calories

Overall, everything goes into calories. If you are in deficit or not overeating, then you will maintain your weight. If overeating compared to what you burn, then you will gain weight. Very simple. The rest is about how and the nuances.

Strength and Cardio Training

Both matter. Strength training builds muscle, bone density, and long-term health. There is more evidence that strength training is more beneficial than cardio. At least you will be stronger and will have healthier older ages in terms of servicing yourself. It is known that most problems of older age are connected to lack of muscle, from simple movement to bones and joints. Cardio protects the heart and keeps the system running. Skipping either is short-sighted.

Fasting

Intermittent fasting is less about magic fat burning and more about behavior. It gives structure, reduces mindless snacking, and helps with calorie control. For me, it is a framework to avoid grazing all day. Too many “experts” speak about fasting and autophagy, when your cells clean the useless ones. The truth and the science is still about overall calorie deficit. If you are in deficit, it will happen in any case, whether fasting or just eating less. The thing is that with fasting, some people find it easier to maintain food intake because eating time is shorter.

Creatine

One of the rare supplements backed by decades of research. Safe, cheap, and proven to support strength and muscle mass. There is also some evidence about cognitive and brain benefits. I take it daily, no mystery here.

Protein Diet

The one clear principle across all research: sufficient protein intake is essential. Protein helps preserve muscle, supports recovery, and delays hunger. When you consume more protein, you stay full longer and it becomes easier to control food intake. A high-protein diet is not a hack, it is the foundation for everything. I came to the conclusion of around 2 grams per kilogram of lean body weight per day.

I have more about this topic here: Protein: The Missing Link in How We Age, Grow, and Stay Strong.

Other Diets

Keto, paleo, vegan, carnivore, every diet has its fans. Most work for the same reason: they create a calorie structure, limit choices, and reduce overconsumption. I don’t follow a strict label diet. I focus on protein, whole foods, and balance. It is very personal, and evidence shows that any kind of branded diet is just about structuring calories. Everyone picks the one that is easiest to follow and fits their daily rhythm.

Supplements

Creatine is in. Protein powder is just food in another form. Beyond that, most supplements add very little compared to diet, sleep, and training. I use a few with decent evidence, like magnesium, sometimes electrolytes, and a few just for testing. But I do not expect miracles.

Late-Night Eating

The body does not shut down at night. What matters is total intake, not the clock. The real problem with late-night eating is behavioral. It is when people reach for junk food, not salmon and broccoli.

Junk Food

Ultra-processed food is engineered to over-deliver taste and calories while under-delivering nutrition. The problem is not eating a candy once, it is building a lifestyle around it. I treat it like entertainment. Fine sometimes, dangerous as a default.

Processed Food

Some food processing is good. Freezing, fermenting, or cooking are also forms of processing. They make food safer and often preserve nutrition.

The problem is ultra-processing. This is not about a single ingredient, it is about the entire factory process. Ultra-processing strips out natural nutrients and adds sugar, unhealthy fats, preservatives, colors, and stabilizers. This is the kind of processing that drives overeating and poor health.

The more a food is changed in a factory, the more likely it is designed for taste and shelf life, not for your health. Preservation itself is not the issue. Natural methods like salt, vinegar, or fermentation have been used for centuries and are safe. Even some modern preservatives are harmless. In the end, it all comes down to quantities and dosage.

Magic Foods and “No-No” Foods

There are no magic foods that transform health overnight, and there are no foods that are pure poison. Some foods are more nutrient dense and support health better, others are less beneficial and easier to overeat. The smart choice is not to worship or demonize single foods but to understand their role in the bigger picture of diet and behavior.

Glucose Is Fuel

Despite the hype, the brain and muscles run primarily on glucose. Carbs are not the enemy. Cutting them to zero is unnecessary unless you have a medical condition. Managing quality and timing of carbs matters more than demonizing them.

Sleep

One of the most important parts. Poor sleep increases appetite, reduces insulin sensitivity, and slows recovery. You cannot fix health without fixing sleep. Your body and brain need rest and reset. Without it, training is less effective, diet is harder to control, and long-term health declines. Consistent, high-quality sleep is the base that supports every other part of fitness and health.

My Takeaway

The science shifts, trends change, but the core is stable:

• Lift and move.

• Eat enough protein.

• Use fasting or diet structures to control behavior.

• Keep junk food rare.

• Do not chase miracles anywhere.

• Protect sleep.

Health is not about secrets. It is about consistent behavior. Framework beats fads. Shortcuts and magic solutions do not exist. Personal trainers or doing it yourself, there is no single truth there either. It is a personal choice, not a rule.

References

• Hall KD, et al. Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men. Am J Clin Nutr. 2016.

• Piercy KL, et al. The Physical Activity Guidelines for Americans. JAMA. 2018.

• Patterson RE, Sears DD. Metabolic Effects of Intermittent Fasting. Annu Rev Nutr. 2017.

• Kreider RB, et al. International Society of Sports Nutrition position stand: safety and efficacy of creatine supplementation. J Int Soc Sports Nutr. 2017.

• Phillips SM, Van Loon LJC. Dietary protein for athletes: from requirements to optimum adaptation. J Sports Sci. 2011.

• Maughan RJ, et al. IOC consensus statement: dietary supplements and the high-performance athlete. Br J Sports Med. 2018.

• Kinsey AW, Ormsbee MJ. The health impact of nighttime eating. Nutrients. 2015.

• Hall KD, et al. Ultra-processed diets cause excess calorie intake and weight gain. Cell Metab. 2019.

• Monteiro CA, et al. The UN Decade of Nutrition, the NOVA classification and the trouble with ultra-processing. Public Health Nutr. 2018.

• Feinman RD, Fine EJ. A physiological perspective on low-carbohydrate diets. Metabolism. 2004.

• Walker M. Why We Sleep. Scribner, 2017.

• Gibney MJ. Ultra-processed foods: Definitions and policy issues. Curr Dev Nutr. 2019.