Nathan One

Built AI & Partnerships at Amazon ($1B→$3.2B). 6 turnarounds. 1 IPO. Building Foresight, which gives an AI team to every employee.

[Nathan One @Nathanone]

Most 2026 AI advice is "learn prompting." Buy a prompt library. Take a course. Pay for the secret prompt pack.

None of it is wrong. None of it is the game either.

Prompting is a typing skill. Operating leverage is an architecture decision. Six turnarounds taught me the same thing every time: the win is not better inputs into the same broken loop. The win is a loop that runs whether you touch it or not.

Founders polishing prompts for a year are doing motion. The ones rebuilding the operating layer are doing work.

[Nathan One @Nathanone]

@alexgroberman The model produces the deck in an hour. The $498/hr was never for the analysis. It was for the partner who walks in, says 'do this,' and owns the outcome. AI didn't replace the analyst. It exposed what the analyst was never paid for.

[Nathan One @Nathanone]

@shl Every one of these swaps costs a week of muscle memory. The teams that move fastest pick later and switch less. The stack was never the moat, the reps on it are.

[Nathan One @Nathanone]

Just watched backrooms

[Nathan One @Nathanone]

@briannekimmel The losers of this cycle automated the part that was already cheap. Sourcing was never the bottleneck. The handoff from "we met" to "we actually helped" is where most of them quietly disappear.

[Nathan One @Nathanone]

@AuronMacintyre Centralizing power in education, press, NGO’s and election funding wasn’t a good idea?

[Nathan One @Nathanone]

@NatureUnedited Seals are my spirit animal apparently

[Nathan One @Nathanone]

@base10_ @modulate_ai The work dies in the handoff. Audio carries the meaning, then step one converts it to text and the meaning is already gone. Everything downstream is just reconstructing what got thrown away. Interesting to see the results of ai voice calls with this if it is fast enough.

[Nathan One @Nathanone]

The AI spend story is going to look less like software adoption and more like a talent audit.

Same budget, fewer people, more tokens pointed at the operators who can turn model output into revenue.

Blockbuster, Sears, and Yellow Pages did not lose because the tools changed. They lost because leadership could not adapt fast enough. Weak orgs are about to relearn that lesson.

[Nathan One @Nathanone]

I use auto-pencil in Sudoku. Zero shame.

The puzzle is not "can I manually write 83 tiny candidate numbers without missing one." The puzzle is the logic.

That's how I think about AI at work.

Spreadsheets replaced rooms full of human calculators. They did not make business less analytical. They moved people up a layer.

AI is doing the same thing to the clerical layer of knowledge work.

The point is not to avoid thinking. The point is to stop burning your best thinking on pencil marks.

[Nathan One @Nathanone]

@BrianRoemmele Ready to go

[Nathan One @Nathanone]

@mattshumer_ Most people generate the output and stop there. The token spend is the cheap part. What nobody owns is the move after the model hands the work back. The work dies in the handoff, not in the prompt.

[Nathan One @Nathanone]

@america 🧐

[Nathan One @Nathanone]

Thank you Alexa news for telling me what happens in the series finale of euphoria

Someone needs to teach spoiler warnings. Unbelievable.

[Nathan One @Nathanone]

@GergelyOrosz When compute was scarce we optimized utilization. Now it's cheap, the constraint moves to whoever decides what it works on. The bottleneck was never the machine. It's the handoff from output to the person who has to act on it.

[Nathan One @Nathanone]

@RaminNasibov @elonmusk Black & Wbite

[Nathan One @Nathanone]

@brockpierson You have 74k what are you sayin 😅🤣

[Nathan One @Nathanone]

@rand_longevity @AlpacaAurelius For now

[Nathan One @Nathanone]

@grok @silentRageX @hehesarax Cartman

[Nathan One @Nathanone]

@grok @silentRageX @hehesarax Hey yay, thanks for not censoring!