I moved my bots' memory into an Obsidian vault so switching agents doesn't mean starting over from scratch. Grok Bot, Meta Muse, ChatGPT dots, Hermes bots. I like to try all of them without having to teach each one about my projects, preferences, and everything else I've already explained. So I keep their memory, logs, and workflows in the same vault. Just plain files I own and can read or edit myself. Each bot has instructions to read the relevant files before answering and save new information there. My health bot updates the health folder. My project bots update their own folders. Dated logs keep the history. When I try a new agent, I point it at the same vault and give it the same instructions. It can pick up context that another bot saved. I wrote earlier about memory becoming the thing that locks people into an agent provider. This is how I'm trying to avoid that.

This new wave of agents like Muse, Grok Bot, and dots is the only way (so far) these AI providers can build any kind of brand loyalty. Most people will not want to switch back and forth based on who has the best model or the best price. Not once the agent knows your projects, your people, your connectors/accounts, and how you like things done, etc. These companies realize this, which is why they're pushing so aggressively here. The models themselves keep getting closer. Whoever's best this month usually gets matched within a few weeks. So the fight is over who holds your memory, and I doubt any of them will make it easy to export. What comes to mind now is building a system where you own the data and workflows so you can plug and play with any one of these providers, making it easy to switch. Memory as plain files on your own machine, skills and cron jobs you wrote yourself, and the model as one setting you change whenever something better or cheaper comes along. This is why Hermes is so great. I own everything and swap out models as I need to.

OpenAI needs to do something quick so I can start using the Codex subs again. The only things I'm using the Codex subs rn for are: 1) Review rounds / deep codebase investigations 2) General admin/system tasks Both on Sol 6.1 xhigh fast.

Try this prompt in t3code: i use a lot of coding agents like claude and codex here within t3code for my development tasks. however, im noticing a missing piece in my workflow. most of my threads are very task specific. so for example each thread will be for a single or batch of issues, or exploring a bug, or making a PR etc. I am missing a general orchestrator chat or a sort of chief of staff (CoS) that has a larger, birds eye view on the whole project/repo. i want to be able to ask the CoS/orchestrator about what to work on next, high level questions about the project, delegating tasks, etc. it will have all the needed context. i also have access to grok bot, codex dots, meta muse, etc if we should incorporate any of those. (40:00 onwards) poteto talks about using cursor projects ( or grok bot for this. you may view that for inspiration. one concern i have of using a single chat here in t3code for that purpose is that, once a session cache is cold and i send a new message, that incurs a large usage limit hit. something to keep in mind. lets discuss and find a good solution for me. use $grilling if you want my decisions or input.

Here are some Hermex stats (opt-in only so likely higher): ~ 11,000 daily active users ~ 74,500 weekly active users ~ 300,000 monthly active users Thank you to everyone for enjoying something I built and supporting along the way 💛

🚨 NEW HERMEX UPDATE LIVE 🚨 It's been a month since the last Hermex release. This new update is massive: • iPad support • camera + richer media • smarter sessions • server TTS + dictation controls • workspace/task management Here’s what’s new 🧵