OMem

Your AI agent is clever, but every session starts blank — it can’t see the work going on around you. OMem fixes that. It runs quietly on your machine, turning your real work — email, calendar, meeting notes, and files (Office formats, PDFs, images, HTML, Markdown) — into a plain-Markdown wiki that any agent can query. Capable agents and your real work context, finally connected.

Available on macOS today; Windows is coming. Free for personal use.

Start here

Install & set up

Five guided minutes from a one-line install to your first answer — or let your AI agent install it for you. Get started →

Understand it

Why it’s shaped this way: the wiki as truth, progressive disclosure, deterministic parsing, multi-path retrieval. Read the concepts →

Use it day to day

Ingest mail, files, calendar and Loop; query from an agent; schedule, observe, back up. How-to guides →

Look something up

Every CLI command, every config field, file formats, exit codes. Reference →

What makes it different

Local-first, zero IT

Reads what your OS already has access to — no vendor cloud, no Graph API consent, no IT ticket.

The wiki is yours

Plain Markdown files in a folder you choose. Read it, grep it, edit it, version it. Nothing locks you in.

Agent-agnostic

One CLI; a thin Skill / MCP wrapper over it. Switch agents next year — your memory persists.

Search that's actually good

Keyword search by default; enable qmd for multi-path semantic + cross-language retrieval. Chinese is first-class.

Keeping current

Changelog What shipped in each release, newest first.

Talk to the author

OMem is built by an IT lead working on the front line of enterprise AI adoption. If you’re using OMem, thinking about it for a team, or just want to talk shop, I’d genuinely like to hear from you.

About the author & how to reach me Who builds OMem, why — and how to get in touch.