Ajit Banerjee

Founder & CEO, @TheSageOx

[Lenny Rachitsky @lennysan]

The biggest upheaval to work in our lifetime is here.

Joe Hudson (@FU_joehudson) coaches the research team at @OpenAI (along with @sama and leaders across Apple and Google), and from that front-row seat, has noticed the people doing best in these AI-forward environments aren't winning for the reasons you'd expect.

It's not more knowledge or more effort. Those are the exact two things AI does best.

Instead, they win on "emotional clarity": staying in hard conversations, not turning on themselves (and others) when things get tough, and pushing forward through failure.

In today's incredible guest post, Joe shows you exactly how to build this skill within yourself and your team.

Inside:
+ How AI is changing work: the NBA-ification of teams
+ The new inner game: the "wisdom stack" that will set you apart
+ How to build a team that can metabolize fear

Don't miss this one: https://t.co/qEVoQw8urs

[aviel @aviel]

There’s a strange trap in how cities think about themselves.

Seattle vs. SF. SF vs. New York. This place vs. that place. For many, your identity as a resident becomes about positioning yourself in opposition to somewhere else.

The World Cup is the perfect reminder of how silly that is. The countries of the world literally compete against each other, but what makes it beautiful is how much everyone shares: the same game, the same stakes, the same joy, the same heartbreak, the same belief that where you come from matters… but does not have to limit where you belong.

The best founders think the same way.

Seattle is an amazing place, but your ambition should not fit inside one city.

With that, there could not be a better time to announce that we’ve now officially opened Foundations SF at 142 Minna to all https://t.co/wWURMtCAmS members, right in the heart of San Francisco.

But wait, Foundations is not moving on from Seattle. In fact, we are going deeper: we just took over our entire building in Capitol Hill. 15,000 sqft of startup goodness, right in the heart of the city.

The point is not to turn Seattle founders into SF founders. Nor is it to create a new Foundations community in SF. It’s to make geography irrelevant.

Build from Seattle. Plug into SF. Do the opposite. Move between both. Belong in both. Who cares.

No permission.
No provincialism.
No “visiting someone else’s ecosystem.”

Foundations is heart to heart. Of the cities and of the spirit. And yes, this is a Starship reference. The band… I’m old 😭

Like with the World Cup, we’re about our own universally shared beautiful game. Entrepreneurship.

All that matters is that you build something that matters, and we just gave over 250 of the best Seattle founders a superpower to do more.

You have a home in both cities, even if you’ve never visited SF before.

LFG.

If you’re in SF on the 30th, we’re doing a small, private office-warming during the AI Engineer World Fair.

https://t.co/zo9Yp0fqa3

No promises on getting accepted, and please don’t just show up.

[Ajit Banerjee @tensorport443]

10/ Three stories. One pattern:
🧛 "Put 2 humans on every line"
🧛 "Ship at a startup, heebie-jeebie at Meta-prod"
🧛 "We can't slow down for a SOC2 checklist"
The shudder is never the enemy. It's the map.
Don't reach for the old tool. Build the one the AI era still needs. 🧛

[Ajit Banerjee @tensorport443]

1/ A number from inside an AI-native team last week:
3 engineers.
3 million lines of code.
600,000 of them deleted in the last year.
The lead's reaction wasn't pride. It was a shudder:
"I have never felt this level of cognitive debt before."

AI Vampire Series, part 3. 🧛🧵
#AIVampires #HeebieJeebies

[Ajit Banerjee @tensorport443]

9/ Cognitive debt and quarterly checklists are the same monster in two masks.
Both assume the human is a bottleneck to protect. In an AI-native team, the human is the scarce comprehension layer — and your tools have to feed it, not slow it.

[Ajit Banerjee @tensorport443]

8/ The real move is to invent the AI-speed instrument:
→ continuous, executable controls instead of quarterly checklists
→ compliance as code that runs every commit
→ comprehension tooling that pays down cognitive debt automatically

[Ajit Banerjee @tensorport443]

7/ The heebie-jeebies hand you two fake choices:
→ Slow to the checklist's speed and lose your only advantage
→ Skip it and fly blind
Both are BC-era thinking. Both are wrong.

[Ajit Banerjee @tensorport443]

6/ Read that carefully — That is NOT "compliance doesn't matter."
It's: the checklist was designed for a team that ships at human speed.
A control built to be verified once a quarter cannot govern a team that rewrites 600K lines a year.
The instrument is out of calibration.

[Ajit Banerjee @tensorport443]

5/ And here's where it gets sharp.
The same week, the question came up: should we run the standard SOC2 checklist?
The answer was immediate:
"There is no way we can go through the slowdown of a SOC2 checklist at this speed."

[Ajit Banerjee @tensorport443]

4/ This is where the heebie-jeebies whisper the old instruction:
"Slow down. Document everything. Hold a design review for every change."
It's the same reflex as "put 2 humans on every line" — reaching for the BC-era tool to soothe an AI-era fear.

[Ajit Banerjee @tensorport443]

3/ Deleting 600K lines is the tell.
In the BC era (Before Claude), that scale of deletion took quarters of careful archaeology. Here it happened in the background of shipping.
The machine rewrites faster than you can re-understand.
Solvent in code. Bankrupt in comprehension.

[Ajit Banerjee @tensorport443]

2/ We spent 20 years learning to fear technical debt — the mess you leave in the code.
The AI era invented a new monster: cognitive debt — the mess you leave in your own head.
The code is clean. The codebase moves faster than any human can hold in working memory. That gap is the debt.

[Ajit Banerjee @tensorport443]

10/ Old world: confidence came from how slowly and carefully a human wrote it.
New world: confidence comes from how thoroughly and fast you can interrogate what the machine wrote.
The heebie-jeebies tell you how big that interrogation needs to be.
Listen to them.🧛

[Ajit Banerjee @tensorport443]

1/ The heebie-jeebies aren't a bug in your judgment.
They're a measuring instrument.

Story #2 of the AI Vampire Series 🧛🧵
#AIVampires #HeebieJeebies

[Ajit Banerjee @tensorport443]

9/ That instrument looks like:
→ load + chaos harnesses that probe failure modes at scale
→ behavior specs for every durability guarantee
→ verification that runs as fast as the code was written