Neuronz.ai

English

Français

English

Français

Connect your agent ​

Neuronz.ai is a managed service: there is nothing to host. You connect your harness — the coding agent you use (Claude Code, the Claude desktop app, Oh-My-Pi) — to your Neuronz.ai account once. Each account has one personal workspace that holds all of your memory. From then on the agent is pointed at your stored context on every prompt and saves new knowledge as you work.

Choose your agent ​

  • Claude Code — fully supported. The Neuronz.ai plugin runs the whole memory loop end to end.
  • Claude desktop app — the Code tab runs the same plugin; chats and Cowork tasks (and claude.ai) connect through a custom connector, where Claude uses your memory when you ask it to.
  • Claude mobile app — chats on your phone use the same connector as the desktop app.
  • Oh-My-Pi — fully supported. Oh-My-Pi is an open-source terminal coding agent; Neuronz.ai installs into it as an extension. A few features differ from Claude Code; see harness parity.

What happens once connected ​

On every prompt, the agent is told which of the profile's facts, past work and knowledge match, and reads the ones it needs. In Claude Code and Oh-My-Pi it also saves durable facts and the work it did as the session runs; in the Claude Code terminal app a final capture also runs after the session ends. A chat in the Claude app or claude.ai saves only when you ask it to. Harness parity lists the differences, and the plugin page lists what runs at each point of a session.

Profiles

Your memory is kept in profiles: separate scopes, usually one per project, so one project's context never shows up in another. The agent picks the profile from the directory it runs in — many repos can share one profile, and a new directory gets its own. A chat in the Claude app has no directory, so it uses the profile you choose as your default chat profile. See Core concepts.

Next ​