01
Remember
Your assistant calls remember when something worth keeping happens: a decision, a commitment, a blocker. Or you capture it yourself in one line.
memory layer · model context protocolYour AI forgets.
Your AI forgets.
3ngram remembers.
A persistent memory layer for your AI tools over MCP. Decisions, commitments, and context are captured once, then carried across Claude, ChatGPT, Cursor, and Codex, every session.
No credit card · Works with Claude, ChatGPT, Cursor & more
recall → context
- decision
Ship the pricing page without the Teams tier until seats land.
- commitment
Send Amara the migration runbook by Thursday.
- blocker
Staging deploys blocked on the expired Postgres cert.
Works inside your MCP clients · one shared context · see setup docs ↗
the problem
Every chat starts from zero. You re-explain the project, re-paste the context, and re-make decisions you already made.
Retrieval is the entry point; follow-through is the product.
01·how it works
Remember. Recall. Carried forward.
No new app to live in. No prompts to memorize. Capture the work state once, then let it show up when the next session, teammate, or agent needs the thread.
02
Recall
On later turns, relevant memories surface by meaning, not keywords, inside whichever MCP client you happen to be working in.
03
Carried forward
Open threads follow the work. Commitments resurface until they're closed, and the next session starts already knowing where you left off.
02·what you get
Built for work you're accountable for.
A memory layer you can read, audit, and correct. Never a black box that quietly rewrites what your tools know.
Semantic recall
Memories surface by meaning, not keywords. Ask about the migration and get the decision, the blocker, and who owns it.
Scopes
Work, personal, shared: memories carry a namespace. Deleting a scope never touches the memories that carry it.
Advisory consolidation
Background jobs propose merging near-duplicates. Proposals wait for your review; nothing is rewritten behind your back.
Content-free reporting
Dashboards, logs, and audit events carry counts and metadata, never memory text. What you remember stays yours to read.
Audit trail
Every memory keeps its history: who captured it, when it changed, and what superseded it. Corrections version; they never destroy.
One MCP endpoint
The same mcp.3ngram.ai/mcp URL in every client, authorized in the browser. No per-tool exports, no copy-paste context.
the product
Everything your AI knows, in one place you can read.
Browse and search every memory, watch open commitments and blockers, and see exactly which tool captured what. Content stays yours; dashboards only count it.
Illustrative preview with sample memories.
06 ·open source
Open source. Run the whole thing yourself.
3ngram is open source. The core is Apache-2.0 and self-hostable, so you never have to take our word for how your context is stored, searched, or forgotten. The code is right there.
$ git clone https://github.com/B3dmar/3ngram
$ cd 3ngram
$ docker compose --env-file .env.selfhost \
-f compose.selfhost.yml up -d
✓ 3ngram server listening on :3000Apache-2.0 core
The memory engine, MCP server, API, and SDK are Apache-2.0. Read it, fork it, run it. The hosted service adds proprietary cloud-operations code on top of the same core.
Self-host in minutes
Clone the repo and bring the stack up with docker compose, backed by your own Postgres + pgvector. The hosted service runs the same core.
MCP-first, no lock-in
Self-host or use the managed cloud and speak the same MCP protocol. Your clients don't change, and your data stays portable.
plans ·2 plans · monthly or annual
Start with shared context. Upgrade for follow-through.
Free gives every AI the same place to remember work, with unlimited memories and search. Pro adds a higher usage budget and priority support today; AI answers and digests are coming soon. Teams stays coming soon while shared-context pricing is settled.
Retrieval is the entry point. Paid plans grow into daily follow-through as digests and AI answers ship.
annual · 2 months free
Shared context for every AI: MCP server, core memory, search, prompts, and dashboard.
- Unlimited memories
- Unlimited search
- Full MCP server
- Web dashboard
- Built-in prompts
- Fair-use usage budget
Get started free
No credit card required.
For AI power users: more headroom today, and first access to AI answers and digests as they ship.
- Everything in Free
- Higher usage budget
- Tags and priority support
- AI answers (RAG search)coming soon
- Daily and weekly digestscoming soon
- Source sync integrationscoming soon
Get started with Pro
Cancel any time. The plan reverts to Free.
feature table ·Free, Pro
What each plan includes
| Feature | Free | Pro |
|---|---|---|
| Full MCP server | FreeIncluded | ProIncluded |
| Memories | FreeUnlimited, fair-use budget | ProUnlimited, higher budget |
| Search | FreeUnlimited search | ProAI answers and RAG search, coming soon |
| Weekly digest | FreeNot included | ProComing soon |
| Daily digest | FreeNot included | ProComing soon |
| Web dashboard | FreeIncluded | ProIncluded |
| Prompts | FreeBuilt-in MCP prompts | ProBuilt-in and custom prompts |
| Prompt marketplace | FreeBrowse public workflows | ProSave and reuse workflows |
| Integrations | FreeMCP clients | ProMCP clients; source sync coming soon |
| Bots | FreeNot included | ProPersonal digests and nudges, coming soon |
| Team workspace | FreeNot included | ProNot included |
Want the full product surface? Read the use-case guides
Give your AI a memory that carries work forward.
Claude, ChatGPT, Cursor, and Codex share one memory, so the work you decide in one session is still accountable in the next.