andrwh

day 3, a little home for your orphaned music 🎶

i've always had this idea of some form of 'personal super app'. like 1 app which combines the random apps on your phone that suck for what they do - webview wrappers, rewards programs, etc but i don't actually want to spend time reverse engineering every app's internal api and building it so instead, i'm getting capy to build it. discussed a bit about the mini-app architecture i wanted, decided on react native + heroui native + uniwind (which is a really nice stack). 1 hour ago, i told capy to get the base of the app working fully, and for it to test the app itself with an android vm. it's been working unprompted with 0 guidance - installing an android environment and working autonomously to fix issues/iterate. it's still cooking as i write this. the software builds itself.

@vishyfishy2 this is crazy

Most AI tools treat planning and execution as one move: you prompt, it writes. That feels fast, until it's not. Capy splits the work into two specialized agents because planning is where quality is won or lost. Better specs mean fewer iteration loops and higher quality outputs.

@ryancarson @linear :) x.com/andrwhcom/stat…

The company that actually builds the agent-first code factory is going to be worth hundreds of billions of dollars. No one has cracked it yet. It can't be the model labs because then you're tied to one model. I'm hoping a company like @linear will do this. I'd happily pay thousands of dollars a month for that (+ the token cost). Basically, we need SDLC 2.0 for the agent age. (Also, the right solution can't rely on gh - we need whoever does this to completely replace it as well.)

DLSS5

so excited to finally release what me and the team have spent many sleepless nights over the last six months working on

@justinsunyt @wilsonhou

@wilsonhou coolest designer of all time!!

coolest launch of march!!

my startup uses @capydotai for complex tasks and it's by far better than any other competitor we've tried (at least 5 others)

Omg Multiplayer????!!!! Gotta try it…

x.com/capydotai/stat… introducing @capydotai. the reason i can ship 1,000 features a day and still travel i just tell it what i wanna do, it tests, reviews and then I say LGTM. do that in 1,000 times in parallel. i'm in tokyo drinking sake rn it's pretty good because so far i haven't been fired yet

Introducing Capy Plan, build, test, review and ship self-healing PRs from everywhere your team works

Introducing Capy The agent that plans, builds, tests, and reviews your code - all in the cloud Multiplayer on web, mobile, and Slack

@juristr @capydotai

Another one for the collection - cmux.dev - conductor.build - superset.sh - (what did I forget??) - dmux.ai

@vishyfishy2 @capydotai suika.live cua benchmarkmaxing

but do you give your agents a gaming break??

We got early access to GPT-5.4 and were surprised by the results from testing in different workflows inside Capy. It's faster and smarter in many ways, but we find that it still has issues on longer context workflows.

@lowercaseclub @skydotcs @0xluffy aw shucks

@andrwhcom @skydotcs @0xluffy idk andrew was pretty g too tho

exciting stuff coming soon new face new interface new capabilities see you guys next week

@skydotcs @0xluffy this was all @lowercaseclub !! the goats

@0xluffy ur designer be cookin

told @capydotai to give me gear 2

"The child in me never wants to let go of the child in you."