Pricing
The software is free and MIT-licensed, and it stays that way.
When your team needs it installed, supported and kept running, you pay for that work.
Plans
-
Community
Free
For individuals and teams who run it themselves.
- The full local memory server: 77 MCP tools, knowledge graph, local embeddings
- The team server: personal, team and shared memory, roles, audit history, backup and restore
- Works with Claude Code, Codex CLI, Cursor and other MCP clients
- Install with pip, Homebrew, npm or Docker
- Community help through GitHub issues
-
Team
from $4,800 / year
Monthly: $490, 3-month minimum.
For one team running its own server.
- Support for 1 production and 1 staging deployment, up to 30 users
- Email support: first response within 1 business day for critical issues
- Upgrade guidance for every release
- 2 named contacts
-
Business
from $12,000 / year
For several teams or departments.
- Up to 3 production deployments, up to 150 users
- Private shared channel: first response within 4 support hours for critical issues
- A guided upgrade session every quarter
- Backup and restore check, plus a server health review, twice a year
- 4 named contacts
-
Enterprise
from $30,000 / year
For organizations with security review, procurement or air-gapped requirements.
- Up to 500 users; custom terms above that
- First response within 2 support hours for critical issues, with limited after-hours coverage and response-time credits
- Named contact, guided upgrades, quarterly backup checks and health reviews
- Security questionnaire and architecture review each year
- Help installing without internet access
- Early notice of security advisories
Prices are starting prices. Support hours, severity levels and the exact response targets are in our support policy.
Implementation packages
Fixed scope, fixed price. You pay half at signature and half when the agreed acceptance criteria are met.
-
Team Server Setup
from $3,500
We install the team server on your infrastructure, set up up to 3 teams and 30 users, connect your agents, schedule backups and test a restore, and run a 2-hour training session. You get a runbook for your admins.
Not ready for a team server? Start with a 2-week pilot → -
Department Knowledge & Onboarding
from $7,500
Everything in Team Server Setup for one department. We also import your existing knowledge and build an onboarding course that new hires take through their own AI agent. Managers see their progress.
Available when the onboarding module is released.
-
Enterprise Rollout
from $18,000
A pilot department, then a rollout plan across the organization, with security review and an air-gapped install if you need one.
-
Custom work
from $1,200 per day
Integrations, importers, custom tools, benchmarks on your own data. Quoted before we start.
Every package includes 30 days of Team-level support.
Why pay for something that is free?
- Your data stays on your servers.
- TAM runs on your hardware. It stores memory in local SQLite files and computes embeddings locally. Optional features that call an LLM use the provider you configure, which can be a local model. We never need a copy of your memory to support you.
- No license keys, no usage meters, nothing phoning home.
- Plans are flat yearly prices with a user range, and we rely on your word for the user count.
- You work with the person who wrote it.
- Support comes from the maintainer of total-agent-memory.
- Quality you can check.
- On held-out LoCoMo and LongMemEval-S questions, we found no statistically significant difference between TAM and Mem0 Platform at the same answering model. How we measured it, including the limits →
FAQ
Is the paid version different from the free one?
No. Everyone runs the same MIT-licensed code. A plan pays for support, installation and response commitments.
Do you host it for us?
Not yet. TAM is built to run on your infrastructure. If you can't run a server yourselves, tell us. We may offer single-tenant hosting later.
Do you have SOC 2 or ISO 27001?
No. We share our security document, threat model and the full source code, and on Enterprise plans we answer your security questionnaire.
What happens if we stop paying?
Nothing changes in the software. It keeps running, because it's yours under the MIT License. You lose access to support.
Can we pay by invoice, and in EUR?
Yes. We send annual invoices in USD or EUR, payable within 30 days.
Which AI agents does it work with?
Claude Code, Codex CLI, Cursor and any client that speaks the Model Context Protocol (MCP). The team server serves MCP over HTTPS, and a small bridge script covers clients that only support stdio.
Do you need our LLM provider keys?
No. You configure your own providers, or a local model through Ollama. Saving and searching memory work without any LLM provider, because embeddings run locally.
Not sure which plan fits?
Tell us your team size and where you want to run it, and we'll reply with a recommendation within 2 business days.
How we handle your data: privacy policy.