Building The World's First AI Software Engineer | Cognition’s Scott Wu
Stop pattern-matching the future based on the past. AI capabilities are doubling every few months—from 10 seconds of autonomous work to hours, soon days, then years. Think from first principles: What becomes possible when everyone has an agent that can execute months of work independently? This isn'
1h 5mKey Takeaway
Stop pattern-matching the future based on the past. AI capabilities are doubling every few months—from 10 seconds of autonomous work to hours, soon days, then years. Think from first principles: What becomes possible when everyone has an agent that can execute months of work independently? This isn't incremental change; it's a phase shift from 'survival mode' to 'creative mode' where the only constraint is what you want to create.
Episode Overview
Scott Wu, founder of Cognition AI (makers of Devon), discusses his intensely competitive nature forged through childhood math competitions, how his parents' immigration from China shaped his worldview, and his vision for AI as the ultimate human-computer interface. He explains why humans struggle to grasp exponential progress and predicts that within 5 years, AI agents will transform how we interact with computers—shifting humanity from survival mode to creative mode.
Key Insights
Competition as Core Identity
Wu's competitive drive isn't learned—it's innate. From age 7-8, competing in seventh-grade math competitions, he was 'salty' about losing before he had evidence of being exceptional. His mother told him he was 'the best' before results proved it, and the family literally displayed math trophies instead of family photos. This early identity formation around competition became the foundation for everything he does.
Exponential Progress Blindness
Humans are evolutionarily wired for linear thinking—a good hunt brings a few days of food. But AI operates on exponential curves where a 'good hunt' could yield a thousand years of food. The Merc report showed AI doing 10-20 seconds of human work; now it's hours. This doubling every few months creates a comprehension gap where pattern-matching the past becomes dangerously misleading.
From Survival Mode to Creative Mode
Wu's co-founder frames the AI revolution as humanity's shift from 'survival mode' (like Minecraft survival mode—gathering resources, avoiding threats) to 'creative mode' (unlimited resources, only question is what you want to create). For the first time in history, the constraint won't be capability but imagination—what do you actually want to make happen in the world?
The Software Economics Revolution
Today, software only makes economic sense if used thousands or millions of times because engineering teams are expensive. But when AI can autonomously build software used once—automating a single task, analyzing one dataset, formatting one email—the entire economics of knowledge work transforms. Every repetitive task becomes a software problem solved on-demand.
First Principles Over Pattern Matching
In 99% of situations, predicting the future by pattern-matching the past works perfectly. AI represents the 1% where it's catastrophically wrong. Instead of asking 'what has always been true,' ask 'why can't an agent work for days, weeks, months, or years unassisted?' First principles thinking reveals possibilities that historical patterns obscure.
Notable Quotes
"I mean it's all I do honestly. I don't know. It's the the way even building a company it feels the same. It's just like you're calculating the moves. You're thinking about okay, if you do this and then this happens and then you do that and here are the different moves and you're like calculating out what comes out to success. You know, it's like a it's like a tree search, you know, where you're exploring the different options in the decision tree and you're trying to figure out how to lead to victory. Like that's like the only thing I do in my life."
"I think the thing that is most human obviously is self-expression creativity like having things that you want to make happen in the world and being able to go do those things right and so I mean a lot of how I think about this is basically we want to build the tool that gives everybody the power to go and make things that they want to make in the world."
"We've been spending all this time living in survival mode as a species, you know, and now we're going to be living in creative mode. Minecraft survival mode is where you're like um you know, you're growing food, you're like making sure you're safe from the monsters at night and whatever. Um and like creative mode is just like everything's up to you. You know, you have all the resources at your disposal. If you want something to happen, it'll happen. And the only question for you is like what you want to make happen."
"It's like humans aren't really wired for this, right? Like all of our like inherent like fight or flight response are our kind of like our ability to kind of like measure things to vastly oversimplify if if you're just, you know, fighting out there, you know, foraging for food or whatever it is like, you know, a good hunt will bring you, you know, a couple days worth of food or something. But obviously with the kind of exponential curves that we deal with, you know, the equivalent of a good hunt here could be a thousand years worth of food and and we don't have that intuitive signal in our brains to really understand that."
"The way that folks typically kind of predict the future or think about what happens next is they pattern match based on what they've seen historically and they say, 'Okay, well for 100 years it's always been like this and it's probably safe to assume that it will be.' And 99% of the time that works great, right? Um and in these particular periods where things that actually move and and they're real things that are different, those are the the 1% of times where it truly is different."
Action Items
-
1
Adopt First Principles Thinking for AI
Stop predicting AI's future by pattern-matching the past. Instead, ask: 'If AI can currently do X hours of work autonomously, why couldn't it do days, weeks, or months?' Map out the implications of each timeline increment for your work, industry, and life. Challenge every assumption based on 'how things have always been.'
-
2
Identify Your 'One-Time Software' Opportunities
List repetitive tasks you do that aren't worth building custom software for (formatting emails, analyzing specific datasets, pulling together reports). These are the tasks AI agents will soon automate on-demand. Start experimenting with tools like Devon or Claude to offload these now, building the muscle for the agent-assisted future.
-
3
Shift from Execution to Vision
If AI will soon handle execution (days/months of autonomous work), your competitive advantage becomes knowing what to build, not how to build it. Spend more time clarifying what you want to create and why, less time on implementation details. Ask yourself: 'If I had an agent that could work for a year unassisted, what would I have it build?'
-
4
Build Your 'AGI Readiness' Mindset
Wu predicts most of this transformation happens in 5 years (which in AI time is 'like a century'). Start preparing now: Learn to communicate tasks clearly to AI, experiment with current tools to understand their capabilities and limitations, and identify which skills will remain human-only versus which will be fully automated.