Shaun

Co-Founder @agentintegrator

[Damian Player @damianplayer]

Chamath on VC pattern recognition: “everybody’s bullshitting.”

less than 2% of Series A investors overlapped across every 50 billion dollar company built since Microsoft.

“if anybody looks at you and tells you they know what the fuck they’re doing, they’re lying.”

[Shaun @shaun_devan7]

At 24, Mark Cuban was going into companies explaining the value of owning a PC.

At 68, he says the next big thing is helping companies implement AI agents.

Do you see a pattern?

[Shaun @shaun_devan7]

@vasuman AI could never come up with this

[Shaun @shaun_devan7]

@zetai_mmt @gork lmaooo

[Shaun @shaun_devan7]

We're still hiring!

[Shaun @shaun_devan7]

We're hiring an AI Engineer at @agentintegrator

Our growth has been insane this past summer we have more pipeline than we can handle.

The work:

-Real systems in production at businesses doing 100's of millions a year (manufacturing, legal, e-com, finance)
-Novel problems to solve in fast growing businesses every week
-Full autonomy. You own your work end to end, from architecture to deployment.
-No client calls, no standups about standups.

Who we're looking for:

-An engineer who's built and deployed agents into production
-Strong full-stack engineer with agent experience
-Someone who sees what AI is about to do to every business in the planet and wants to be the one building it

Early seat at a fast growing startup, with real ownership and real upside. Equity in the company and rev share on projects we take on.

To apply, email [email protected] with:

A paragraph on who you are and why we should hire you

CV / resume

GitHub

If you're a good fit you'll have an offer in your hand at the end of the week.

Know someone who'd be a fit? Send them our way. Finders fee if we hire them.

[Shaun @shaun_devan7]

@damianplayer his stanford talks are goldmines

[Damian Player @damianplayer]

Chamath to a room of Stanford MBAs: "get the fucking money. I'm serious."

he says capital is a moral imperative. the money gets made and allocated either way, so if you have a worldview and none of it behind you, your view doesn't compete.

"in the absence of capital, you're irrelevant. with capital, you're powerful."

[Shaun @shaun_devan7]

@sational_sg i'll check it out

[Shaun @shaun_devan7]

@internetDevv I gotcha

[Shaun @shaun_devan7]

bullish on Ramp

[Karim Atiyeh @karimatiyeh]

Most people judge AI models by what they can do today. @rahulgs judges the slope.

When we started Ramp, people had tasted better software at home and could no longer tolerate the tools they were handed at work. So we built something better.

Same thing is happening now with AI. Our customers use these models every day. They know what speed and intelligence feel like — and anything that makes them repeat themselves or wait for an answer feels broken.

Rahul builds for that. Not for the best model today, but for where models will be in 3–6 months. Studying traces. Improving the harness. Cutting CI time. Giving agents real work with real guardrails.

He gets into all of it with @bcherny, @Anthropic's Head of Claude Code, and our staff engineer @austospumanto.

https://t.co/sLQV0Q2VkK

[Shaun @shaun_devan7]

a BILLION!

[Damian Player @damianplayer]

another billion people started using AI since February.

your mom uses it to write emails and thinks it's part of Google.

most of them will NEVER open a terminal. crazy when you think about it…

[Shaun @shaun_devan7]

@internetDevv 100%

[Shaun @shaun_devan7]

Every company I talk to has an AI agent demo.

Almost none of them have an agent in production.

I've spent the last 6 months deploying agents into real businesses & I can tell you the gap has nothing to do with the model.

The model is the easy part.

Here's what the agents that actually survive production look like:

1) They're built for the ugly stuff. A demo works because someone fed it a clean input in a conference room. Production is missing data, formats you don't control, & records that don't exist. The agents that live are the ones where somebody sat down & listed every way the task goes wrong BEFORE it went live.

2) They're boring. 90% deterministic software & API calls (read: regular code), 10% model. The software is predictable... same input, same output, testable. The model is the one piece that can be wrong in ways you can't predict, so it gets the smallest job in the building: the judgment calls a person used to make.

3) They have a hard NO list. Email a customer, commit an order, move money... none of it happens without a human clicking approve. Day one, everything is gated. The training wheels come off as the agent earns it.

4) They keep receipts. Every single action logged from day one. When something breaks (it will), you trace exactly what the agent did & why. You can't hold an agent accountable if you can't see what it did.

5) They start embarrassingly small. One workflow, return in 3-5 weeks. Then you build the agent layer... the shared foundation every future agent sits on & pulls context from. One-off agents are how companies end up with 15 disconnected pilots & nothing in production.

The companies winning with agents aren't the ones with the best models.

They're the ones treating this like software engineering, because it mostly is.

[Shaun @shaun_devan7]

Every company has an agent demo. Almost none have an agent in production.

I've spent the last 6 months deploying agents into real businesses & the gap has nothing to do with the model.

The agents that survive production:

- 90% regular software & API calls, 10% model judgment (the calls a person used to make)
- built for the ugly inputs: missing data, formats you don't control, records that don't exist
- a hard NO list: email a customer, commit an order, move money... nothing happens without human approval
- every action logged from day one, so you can trace exactly what it did & why
- started embarrassingly small: one workflow, return in 3-5 weeks, then the shared agent layer for everything after

It's mostly software engineering.

[Shaun @shaun_devan7]

@codyschneider 100% - production agents are 90% software and a model making a judgement or api call 10% of the time

[Shaun @shaun_devan7]

Peter Thiel on Steve Jobs' brilliance in long term planning:

[Shaun @shaun_devan7]

@ayenisholah @agentintegrator yep!

[Shaun @shaun_devan7]

We're hiring an AI Engineer at @agentintegrator

Our growth has been insane this past summer we have more pipeline than we can handle.

The work:

-Real systems in production at businesses doing 100's of millions a year (manufacturing, legal, e-com, finance)
-Novel problems to solve in fast growing businesses every week
-Full autonomy. You own your work end to end, from architecture to deployment.
-No client calls, no standups about standups.

Who we're looking for:

-An engineer who's built and deployed agents into production
-Strong full-stack engineer with agent experience
-Someone who sees what AI is about to do to every business in the planet and wants to be the one building it

Early seat at a fast growing startup, with real ownership and real upside. Equity in the company and rev share on projects we take on.

To apply, email [email protected] with:

A paragraph on who you are and why we should hire you

CV / resume

GitHub

If you're a good fit you'll have an offer in your hand at the end of the week.

Know someone who'd be a fit? Send them our way. Finders fee if we hire them.

[Shaun @shaun_devan7]

@SohaibAnwa11 @agentintegrator That'd be great Sohaib - dm me!

[Damian Player @damianplayer]

we are living in the jack-of-all-trades era. a generalist with AI tools is replacing the specialist who only knows one thing. depth used to win.

now it's range + leverage.

[Shaun @shaun_devan7]

bad day to be granola

[Tanay Kothari @tankots]

It's no secret how we feel about typing at Wispr Flow.

The truth is, we hate messy meeting notes too.

Introducing Notetaker 👇

https://t.co/NMvER57a5w