memU

memU gives connected AI agents a shared memory of preferences, project decisions and learned workflows. It's for people who move between agents or sessions and want useful context to carry into later tasks.

• Shared recall: Knowledge captured in one connected agent can be retrieved by another, including across devices that use the same memory backend.
• Inspectable records: Memory stays readable as Markdown, with source links that let you check where a remembered preference or decision came from.
• Reusable skills: Connected agents can turn session history into reusable workflows, update existing skills or skip details that don't deserve a lasting record.
• Agent integrations: Host adapters support Codex, Claude Code, Cursor, OpenClaw, Hermes and WorkBuddy, with availability and limitations varying by operating system.

Host integrations run on macOS, Windows and Linux. Ordinary ChatGPT chat and Claude chat aren't supported; the Claude integration is for Claude Code.

Local deployment lets you choose your own storage and embedding provider, with SQLite or PostgreSQL plus pgvector as storage options. The free hosted service uses a cloud memory backend and requires an account and API key. Supported private deployments are also available through enterprise engagements. The connected agent decides what to remember; memU stores, embeds and retrieves records without making its own LLM calls.