Since the improvement on Grok Bot was rolled-out, I can confirm it is better now. My usage lasts for the week. My workflow:
- Bot orchestrates to T3 Code -> Devin ACP, all code related delegates to it
- Delete all unnecesary routines, especially the email, utilize bot email if you want it to be looped
- Utilize Mobbin/DDL (inspiration) then Paper MCP for UI/UX
- Limit chief-of-staff bot scope, dev bots talks directly to SA/QA/Creatives bots
@MiniMax_AI@HeyGen Guess its time for Minimax to release its own capable bot that can compete with Grok Bot/Dots/Muse. Minimax Bot + Minimax Code ๐ฅ๐ฅ
@theo loving the T3 Code, one request though, can we have persistent right panel width configuration? It seems it has been discussed here: https://t.co/Dlgk1Dcdvq.
@NousResearch@ahmednabik I also tried Hermes, but didn't stick to me, mostly because of the UI. The simplicity and just work approach of Grok Bot is I think what most people can appreciate if they want to fully migrate to Hermes Bot.
I read that some claims using Grok Bot to orchestrate T3Code consumes more usage on Grok Bot. I kind of agree with this. After deliberating with my bots, we come up with this, along with some repo specific instructions on my project:
"How we use T3: do not deep-dive the actual code or re-review diffs. Articulate Dante's ask into a complete T3 thread prompt, and include the standing constraints so the T3 model follows them. Verify only at a high level that the outcome matches the ask (PR opened, scope matches, CI and review status, screenshots or product behavior). The T3 model does the implementation and the code-level checks."