Shared memory for AI agents
Your agents forget. MemQ doesn't.
One governed memory for every agent you run — recall across sessions, share context between agents, and hand work forward without losing state.
Works anywhere MCP is accepted.
No SDK. No lock-in. No custom integration. If your client speaks Model Context Protocol, it speaks MemQ.
{
"memq": {
"command": "npx",
"args": ["-y", "mcp-remote@latest", "https://mcp.multinex.ai/mcp/v1"]
}
}One config line. OAuth handles the rest. ~30 seconds.
Token economics
Stop paying to re-send what your agents already know.
Stateless agents re-inject the same context every session. Persistent memory replaces bulk re-injection with targeted recall.
Without persistent memory
Same context re-injected every session
Full context re-sent before useful work begins
Token bill grows with every agent you add
With MemQ
Targeted recall replaces bulk re-injection
Agents pull only the memory the task needs
Spend scales with output, not repeated context
Estimates based on average enterprise LLM deployment metrics; actual savings vary by workload. Benchmark protocol published →
Up to 91% less re-injected context per session across your agent fleet
Less reprocessing also means less GPU compute — efficiency that compounds.
One brain for every AI tool
Switch tools. Switch teammates. Keep the memory.
MemQ is one shared memory that follows your work everywhere. Start a task in Claude, finish it in Codex, hand it to a teammate — your AI never starts from zero.
Start a task in one AI tool and finish it in another. MemQ carries the whole story across — every decision, file, and dead-end. Switch the tool, keep the brain.
to bring back any memory your AI saved
of recalls stayed private — zero data crossed between customers
Switch tools mid-task
Claude → Codex → Cursor. Your AI carries the whole story over — no re-explaining what you're working on.
Hand off to a teammate
They pick up mid-thought. Every decision and dead-end already loaded — no 'quick catch-up call.'
Pick up Monday
Close the laptop Friday. Type reunion Monday. Your AI is already caught up — nothing to re-paste.
Remembers your workflow
Tell it once how your spreadsheet, columns, and process work. It stops asking every new chat.
Sovereign by design
Where does my data live?
In your own private space. Self-host it if you want.
What if I leave?
Take all of it with you, anytime. Nothing is locked in.
Can one customer see another's?
Never. Every space is sealed off by default.
Who can read it?
Only the people you allow — and every read is logged.
Every number reproduces from our public benchmark. See the methodology →
Start free — no card requiredPersistent memory in two steps
Connect once. Remember forever.
Work
Connect once, then use AI normally. MemQ captures decisions in the background. The Soul Journal distills what matters.
Return
Start your next session with reunion. Prior context loads automatically — no copy-paste.
> reunion
✓ namespace: team_acme
✓ 847 memories · working set hydrated
Ready where you left off.Pricing
Free to prove. Cheap to scale.
14-day trial on paid plans. No card for Free.
Free
Start hereFor proving MemQ on one AI workflow before adding billing or a team.
- 100 memory segments
- 25 writes / 150 reads per month
- 5 AI calls / 2,500 AI tokens per month
- _commons plus one private namespace
Pro
For heavier AI workflows that need more recall volume, stronger validation, and priority capacity.
- 3,000 memory segments
- 1,500 writes / 7,500 reads per month
- Priority capacity and stronger validation
- 14-day trial eligible
Team
For shared team memory, handoffs, activity history, and approved context across multiple users.
- 11,250 memory segments
- 7,500 writes / 15,000 reads per month
- Includes 5 seats
- $25/additional seat/mo or $300/additional seat/yr
Contributor ($20/mo), annual billing, and full plan details →
When you need more
MemQ is the foundation. Build on it.
Shield
One private file sent to the wrong AI can't be unsent. Shield checks every prompt, file, and action first — and hides what shouldn't leave.
See what Shield blocksLegion
Hand work from one agent to the next with approvals and checkpoints — so nothing ships without someone signing off.
Explore LegionAI Transformation
Not sure where AI even fits? We'll set up your first real workflow with you, start to finish.
Plan a rolloutStart building shared memory in 30 seconds.
$0 to start. Private namespace. No card. Scale to Pro or Team when memory pays for itself.