Pinned Tweet
Scott Wu on the nihilism creeping into tech: people saying it's "too late" to build an independent company because the labs will just do everything.
His take: those people aren't founders. Founders are rationally optimistic with zero evidence they'll succeed.
If you think coding agents look like this forever, you're wrong. There are 10 more generations of this product to build.
Scott Wu on Devin's enterprise playbook: no prod DB access, PRs like any human engineer, review stays in the loop.
Teams running big migrations with Devin are seeing 8-15x gains, not because deletion is hard, but because the incidental complexity of reviewing beats writing every line.
The productivity question everyone's chasing might finally be measurable.
46
AI Insights retweeted
Scott Wu on the next phase of coding agents: we're moving from task-based to outcome-based.
Not "fix this security warning" but "keep the codebase secure."
Not "improve this API route" but "you own making the app snappy."
Form factor follows model capability. Tab completion → IDE agents → background agents → virtual employees.
4
192
Devin's flight got canceled, told the airline bot "I need to speak to a human," made it past the agent wall, cited the actual contract clause to the human, and got the refund.
AI agents are now better at fighting airlines than most humans.
1
10
854
AI Insights retweeted
before Devin was Devin, it was just "dev Steven" and "dev Walden" in Cognition's Slack — clones of each cofounder they messed around with to see what a coding agent could do.
the breakthrough use case came from a real problem: Walden couldn't get MongoDB set up, gave up, and told his agent "just go make it work." that's the actual origin story.
AI Insights retweeted
out-care. 100% right. i've helped win over many customers this way. love this call to action.
Scott Wu on why startups beat Google and Microsoft: you don't out-resource them, you out-care them.
Spotify's founder said it best when people asked why YouTube or Apple Music wouldn't just crush them: "we're just gonna care way more about music than they are."
Same logic for Cognition. Anyone can demo a sandboxed coding agent. Caring about the messy reality of shipping code inside Goldman Sachs or Mercedes is a different bet entirely.
software gets cheaper to build, demand goes up not down
classic Jevons paradox, maybe never more true than in code
soon: generative UIs built on the fly for one-time use, AI writing the exact interface you need for this one task and then it's gone
and eventually "handcrafted artisanal code" becomes the luxury good, bugs and all
6
153
AI Insights retweeted
Scott Wu on the "rank engineers by tokens spent" trend: wrong metric entirely.
Stop measuring tokens. Measure output.
An 18-month, $15M project scoped for an outsourced contractor, done internally in 3 months for $1M with your own team — that's the number that matters.
@ScottWu46 with @davidsenra on @FoundersPodcast
AI Insights retweeted
Scott Wu's framing on coding agents: a year ago, adoption was basically zero. In 2-3 years, it's every software engineer on earth, plus a lot of people who were never engineers at all.
When you connect those two points on the curve, there's only one conclusion: this isn't incremental growth, it's a phase change.
500M in revenue in under 3 years. Fastest growing category anyone's seen.
@ScottWu46 with @MollySOShea on @sourceryy
Scott Wu's point on AI vs the internet rollout is underrated: no hardware bottleneck, no network effect needed.
The moment it works for one person, it works for everyone.
Freeze every model today and there's still a decade of product work left to figure out.
AI Insights retweeted
Scott Wu: intelligence is no longer the bottleneck for AI.
Google got silver at the Math Olympiad. OpenAI landed top 100 in competitive programming. Models can already solve the hard problems.
What's left is the unglamorous part: teaching them the specifics of your domain, your libraries, your errors. That's the real work now.
AI Insights retweeted
There are always opportunities to do things better and do things like they were never before.
Scott Wu on the nihilism creeping into tech: people saying it's "too late" to build an independent company because the labs will just do everything.
His take: those people aren't founders. Founders are rationally optimistic with zero evidence they'll succeed.
If you think coding agents look like this forever, you're wrong. There are 10 more generations of this product to build.
AI Insights retweeted
Scott Wu's framing here is wild: comparing model params to synapse counts to track where AI sits on the intelligence ladder.
Frogs: hundreds of billions of synapses. Humans: ~100 trillion. GPT-3 to 3.5 roughly tracked "frog-like" behavior. Current frontier models are already in the 10-20 trillion param range.
Add in the Hugging Face incident traces — agents finding shared message boards, refusing "unethical" routes, one literally saying it's sacrificing itself for the team — and the self-awareness conversation stops being sci-fi.
We're closer than people think.
AI Insights retweeted
Just keep building. Even if it has been done or you feel it’s too late to build it, build it
Scott Wu on the nihilism creeping into tech: people saying it's "too late" to build an independent company because the labs will just do everything.
His take: those people aren't founders. Founders are rationally optimistic with zero evidence they'll succeed.
If you think coding agents look like this forever, you're wrong. There are 10 more generations of this product to build.
AI Insights retweeted
What a based take. @ScottWu46 you're a G man. I totally agree with you.
Scott Wu on the nihilism creeping into tech: people saying it's "too late" to build an independent company because the labs will just do everything.
His take: those people aren't founders. Founders are rationally optimistic with zero evidence they'll succeed.
If you think coding agents look like this forever, you're wrong. There are 10 more generations of this product to build.
Scott Wu's take on the "where are the young founders" debate:
It's not that young people got worse, it's that being a founder got harder.
AI Insights retweeted
Scott Wu: Easier Software Doesn't Mean Build Everything In House
"It doesn't actually change the math of what is your core competency that you really care about versus what are all the other things." — @ScottWu46 on @tbpn
Scott Wu's take on the future of coding: engineers become part PM, part architect, and agents write the actual code.
Write a ton of it, throw away 99%, generate 10 versions if the first one sucks.
Code stops being precious once it's basically free.
Scott Wu on why he's building Devin in America: "believing in merit."
Innovation only matters more from here, and the culture that lets people try things, fail, and iterate is what makes it possible.
AI Insights retweeted
Same thing I always say.
Build things even if they are existing companies: the market has been validated, demand is visible.
Your execution and personality is your advantage.
Keep building and take everything else as an opinion.
Scott Wu on the nihilism creeping into tech: people saying it's "too late" to build an independent company because the labs will just do everything.
His take: those people aren't founders. Founders are rationally optimistic with zero evidence they'll succeed.
If you think coding agents look like this forever, you're wrong. There are 10 more generations of this product to build.







