Hi Everyone. I'm the maker of Verzeta Studio. It started with using several models side by side and noticing each was good at something different. A small local model was fast and fine for reading files and routine work; a bigger cloud model was better at judgement calls but too expensive to use for everything. Every app made me pick one per conversation, so I kept copying text between windows. And I didn't want a cloud account between me and my own files. So Verzeta gives you a team instead of one assistant. You name the members, give each its own provider, model, system prompt and tools, and they work in the same thread, handing work to each other by name. You stay in the conversation and can redirect any of them while they work. The agents can read and write files in a project folder, run shell commands against an allow-list you control, search the web, write and run code in a side canvas, work through task plans, open a poll when they disagree, and run on a schedule while you're away. Every action lands in a timeline you can audit later. Nothing to code: agents are set up in the UI, and project rooms come from templates. Everything runs on your machine. No account, no telemetry, no cloud. It works with OpenAI, Anthropic, Gemini, OpenRouter, DeepSeek, Ollama, llama.cpp and any OpenAI-compatible server, and one build has llama.cpp compiled in. It's a native Qt 6 app with KDE Kirigami, not Electron, with small Android and VS Code clients that pair over your own network. Open Source, Explore Verzeta: https://github.com/Verzeta/

Verzeta Studio

Your AI team, running on your machine, in one chat.

Your AI team, running on your machine, in one chat.