Jan Moehlmann

founder @ComposalAI (YC F26). your agents build it, ours break it

[Jan Moehlmann @JanMoehlmann]

wondering what you are about to ship🔎

[Jan Moehlmann @JanMoehlmann]

wise words as always.. test it like your users would @ComposalAI

[Garry Tan @garrytan]

Something changed in the latest frontier models where cross model review and code review is just less useful now

[Jan Moehlmann @JanMoehlmann]

the next 5 years will be wild!

[Patrick Collison @patrickc]

I'm now building an "improvements" feature into all of my personal vibecoded applications: a freeform text input that's hooked up to an agent that yolo implements whatever is submitted.

It's an interesting experience -- the activation energy barrier to sanding any rough edge or implementing any new feature becomes extremely low. Someone else who uses one of the applications I created has already implemented 150 improvements over the course of a few days. This pattern feels like a new kind of tool -- liquid software? -- and every application that doesn't have this feature now has a slightly rigid and anachronistic flavor. Why can't I just tell it how I want it to be different?

Now, it's not obvious to me how this pattern can or should scale. Does it become unmaintainable after a while? (Maybe model intelligence improvement outstrips my feature appetite such that it's fine?) How does collaborative software work in this model? Can software designed this way be secure and reliable?

Whatever the larger questions about broader deployment, it definitely works great in a personal context. It reminds me a little of the visions of Lisp/Lisp machines, Smalltalk, and HyperCard -- systems that the user can take control of and modify. Those visions never quite worked, though: the vast majority of software today is not end-user modifiable in any useful way. So, while this "liquid" affordance is a very small touch, the way it changes one's relation to the software feels a bit transgressive and empowering in a Lickliderian and Engelbartian way.

[Jan Moehlmann @JanMoehlmann]

879 PRs last week 🤯 @PandelisZ did 548 of them

secret sauce: @greptile + @ComposalAI so it actually works

[Jan Moehlmann @JanMoehlmann]

@mariagorskikh @SajKhosa @KhraminaDianka @ParamMehta24 looks like a solid crew

[Jan Moehlmann @JanMoehlmann]

Can't disagree with that one...

[Y Combinator @ycombinator]

Consider this a sign.

https://t.co/OnqI05W53Z

[Jan Moehlmann @JanMoehlmann]

@AnthropicAI and @OpenAI models get weirdly cautious when you ask them to hunt bugs. @deepseek_ai flash on max just goes for it 🐛

[Jan Moehlmann @JanMoehlmann]

@jonathandavies Great stuff

[Jan Moehlmann @JanMoehlmann]

agents made code a commodity but knowing it actually works is 💀

the agent said it was done... we found 91 issues 🤯

[Jan Moehlmann @JanMoehlmann]

another epic week in AI land..

@AnthropicAI drops its S1 and @OpenAI & @instinct rumoured raising at $1.25tn and $30bn 🤯

[Jan Moehlmann @JanMoehlmann]

@Bartonaut55 under water there is pressure

[Jan Moehlmann @JanMoehlmann]

One day since launch & already living the high life

[Jan Moehlmann @JanMoehlmann]

@onethomaschan @JustinTang29 @LanternAI__ @ycombinator Looks great. Will try it out !

[Jan Moehlmann @JanMoehlmann]

@jonathandavies @exedev @sentry especially when vibe coding stuff - directly to prod

[Jan Moehlmann @JanMoehlmann]

let's do this! 🚀🚀

[Jan Moehlmann @JanMoehlmann]

@MatthewBerman Amazing video! LFG!

[Jan Moehlmann @JanMoehlmann]

@micmakarov Amazing! LFG!