What Pokémon Teaches Us About Building With AI

Key Takeaways

It's funny like at squint value, Pokemon couldn't be more different than writing code or or all the other stuff that people do at Quad. Um, but this core thing like the ability to come up with a plan that's good to try something to see if it's working or not and adjust to understand like what are the different strategies available to me and to sort of like be willing to try them fail update what you should be doing with new information. That is like the core recipe of I think like a lot of what makes agents good in a lot of scenarios. So one that we probably spend too much time talking about is coding. But like when you are writing code, you write something, you see a test fail and you have to think like what did I do wrong? What should I try next? What are the what's the next strategy? And models do this all the time. And there's like one thing between a model that can get it perfectly right every time all the time. But sometimes you just don't even have the information you need. like you don't know until you run the test that you that something you miss something and so the ability to know what test should I run to learn when I find out something how should I incorporate that how should I update the strategy I had to go write this piece of code or whatever that's the same thing and I think that's actually like any industry all of that building a strategy and understanding how to like incorporate new information over potentially like a lot of different actions just has like really broadreaching applications I think for how people build things with quad and it just It maps really intuitively to how we as people think about solving complex problems.