I'm building OpenRoutine, a routine app with a small friend who keeps you company through each step.
No account, no sign-up. Your routines live on your phone, or in your own Google Drive if you want a backup.
first the plan was only their own models. now they're opening up to everyone else's because theirs aren't the best at anything. maybe grok ends up being a harness
Important note regarding Grok @Bot:
Going forward, @SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs.
Whatever is most likely to give you the best outcome.
3 minutes. that's how long it took a bot to find my fresh dokploy panel, claim it and get root. i was literally buying a bottle of water. let me drink my water in peace bro
i don't do tdd, i do spec-driven. an agent and i write the spec together, so the weird edge cases come out early and i get files i can check later. once it's built, two qa agents from different tools review it, then another merges, waits for the deploy and tests it on dev
today claude code built a feature, then asked cursor and deepseek in opencode to review it and compared what they found. all through herdr, all sharing the same project memory. been loving it
@leerob@bot it should make really good designs and have a better memory, claude code remember more things than my grok bot and also it should better at research/improving cycle
@hagen_hu still think most apps don't need to go native. agents can write both codebases, sure, but why burn double the tokens unless your use case actually needs it