←
Peter Pistorius17,694 posts
Peter Pistorius avatar

Peter Pistorius

@appfactory

AI Builder at "startup in stealth." Previously built @RedwoodJS, local-ci.dev, cmdcmd, machinen.dev, and @_snaplet

◷ Joined April 2009▧ 826 Photos and videos
hannah avatar
⟳ Peter Pistorius retweeted
hannah@erhannah·Oct 18

one day you wrote console.log("here") for the last time and didn't even know it

Peter Pistorius avatar
Peter Pistorius@appfactory·Oct 19

The only way I think about computers now is that they have to be glued together; and I don't want to see seams.

Tim Sweeney avatar
⟳ Peter Pistorius retweeted
Tim Sweeney@TimSweeneyEpic·Oct 17

The games we've long wanted, which didn't exist because they'd cost several times more to make than they'd earn back, will soon become possible. As will the games developers wanted to build, but couldn't with the resources available. It'll be a new golden age of gaming.

Peter Pistorius avatar
Peter Pistorius@appfactory·Oct 15

I really like @t3dotcodes, but I wanted it to be more like an inbox: where the moment I respond a thread go away, and only comes back when it needs me. I also run many threads over many projects but they're the same concept, so having #tags was helpful.

Peter Pistorius avatar
Peter Pistorius@appfactory·Oct 15

I built my own agent orchestrator. I've been measuring my "attention" in order to figure out what makes me respond faster.

Post attachment
Kenton Varda avatar
⟳ Peter Pistorius retweeted
Kenton Varda@KentonVarda·Oct 9

This isn't your typical acquisition blog post. I'm really excited not just that @rough__sea, @piscisaureus, and friends are joining, but also what they plan to work on: Making workerd a first-class, self-hostable open source runtime. Read for more: blog.cloudflare.com/deno-joi…

Peter Pistorius avatar
Peter Pistorius@appfactory·Oct 9

Am I qualified to have this opinion? Yes! I have used a ton of them for several weeks at a time. And I am building my own with my own research ideas, measurements of outcomes, and desires.

Peter Pistorius avatar
Peter Pistorius@appfactory·Oct 9

This is a popular opinion now, but T3 Code is a pleasure to use. It is the most sane version of an agent orchestrator that I've used: It has well thought out defaults and good performance. Well done T3 Team!

Peter Pistorius avatar
Peter Pistorius@appfactory·Oct 8

This time last year was React Conf 2025. I remember some of the speakers wondering why there was "so much AI" at a React Conf. This feels like a lifetime ago.

Post attachment
Andrew Jefferson avatar
⟳ Peter Pistorius retweeted
Andrew Jefferson@EastlondonDev·Oct 8

Same has always been true about teams of software engineers. I’ve spent years telling teams to write fewer tests especially unit tests and focus on velocity + minimum necessary end to end and integration testing

ok it's official... AI-written unit tests and integration tests are empirically proven to be unhelpful. you should tell your agents to stop writing tests by themselves on the deepswe eval set, banning sonnet 5.5 high from writing any tests actually resulted in slightly higher success rate (non stat-sig), with less time and token spent (stat-sig). this is pretty hard empirical evidence across the tests that were written by the "tests allowed" baseline arm, 65% of them were unit tests, 35% were integration tests, neither bucket resulted in any improvement compared to not writing any tests at all i also picked a random sample of 44 tasks subset where i completely disabled executing even existing tests - it also did not affect success rate at all i spot checked many tests written in the baseline arm, and my intuition is that most tests are simply a repetition of the implementation agent-written tests do not add any value because both the implementation and the tests were simply the agent's interpretation of our intent. the tests aren't any more accurate than the implementation itself i suspect we can still extract some value from unit tests and integration tests if we describe them ourselves when we believe we can articulate our intent better through test cases than through requirements. but i have not proven this yet also worth noting, during deepswe eval the agent wrote almost no e2e tests (only 17, compared to 3000+ unit/integration tests written). so this analysis does not prove nor disprove the value of e2e tests - i will do another eval specifically for that

Post attachment