Mimzie, an agent-neutral alternative to the Claude Suite/Claude Cowork

hi HN!I built Mimzie, an agent neutral AI workspace inspired by the Claude Suite (Docs, Sheets, Slides) and Cowork.It's a free desktop app that runs ChatGPT, Claude, Gemini, and Grok in their native CLI harnesses. Think T3 Code or Conductor but for knowledge work. It's basically a GUI for managing knowledge work agents in various form factors. One of those form factors is Bots, akin to Grok Bot or Dots.The original idea for mimzie was simple: I pay for Claude and ChatGPT, and a lot of my day-to-day work lives in .xlsx, .pptx and .docx files. Moreover, I was sick and tired of losing context between AI platforms. I didn't want to bet the house on one AI lab's product, be it the ChatGPT or the Claude Desktop app.So I built an AI desktop app that was agent neutral.The goal of mimzie is to bring the kind of agent workflows that devs have had with codex and claude code into the hands of everyone. And to have a friendly GUI to do it. I've found that the personal agents, hermes and openclaw, are intimidating to set up and confusing to the average non-technical person.some of the core principles of mimzie:- your file stays your file: edits are done locally and .docx, .pptx, .xlsx, and .md and synced to the cloud after - every change is reviewable: edits show up as tracked diffs that you accept or reject, at the smallest unit that makes sense (e.g. a sentence in .docx) - free, and no new subscription needed: bring whichever AI plan or plans you already have. You shouldn't have to pay twice for the same inference, and you SHOULD take advantage of the insane subsidization in the current ChatGPT and Claude plans. - Bots with any AI: I love the concept of grok bot, not so much its limited choice of models. I strongly believe that long-standing "ai teammates" should be agent neutral and in your controlTry mimzie out: it's a free beta at mimzie.comit's early, lots of rough edges, but I've had the time of my life developing it.I'd love to answer questions and hear any and all feedback.