Follow-through layer for AI-native work

Keep AI work moving after the chat ends.

Capture commitments, decisions, and blockers from Claude, ChatGPT, and Cursor, then turn them into daily execution through digests, reminders, and reviews.

Already have an account? Sign in

Works withClaude·ChatGPT·Cursor·Windsurf
Claude.ai
connected

You

“Remember that my top priority this week is finishing the Q2 planning doc.”

3ngram

savedmemory #2041 · topic: planning
3ngram MCP·semantic memory
The problem

Your AI is brilliant - and has the memory of a goldfish.

You make commitments in AI conversations. You set intentions. You make decisions. And all of it evaporates when the session ends. The smarter your AI gets, the worse this gets - you trust it with bigger work, but nobody is making sure things actually land.

Native AI memory remembers you prefer Python.

3ngram reminds you that you promised a deliverable by Friday.

How it works

Three steps to a Chief of Staff.

Connect

Add 3ngram to Claude, ChatGPT, Cursor, or any MCP client. Works inside the tools you already use.

Capture

Save commitments, decisions, and context as you work. 3ngram structures it automatically.

Follow through

Get briefings, digests, and alerts that surface what needs attention. Nothing slips through.

In conversation

Your Chief of Staff is always listening.

Save context mid-conversation and pick it back up later - even in a different AI tool. 3ngram keeps the work moving after the chat scrolls away.

Claude.ai
connected

You

“Remember that my top priority this week is finishing the Q2 planning doc.”

3ngram

savedmemory #2041 · topic: planning

You

“What did I say my priority was?”

3ngram

Your top priority this week is finishing the Q2 planning doc. (saved 2h ago)

3ngram MCP·37 tools·semantic memory
In your editor

Follow-through, not just memory.

Decisions, commitments, and blockers stay structured wherever you work. 3ngram tracks due dates, flags overdue items, and sends digests. Your AI doesn't just answer questions - it tells you what you should be working on.

~/projects/api · claude
connected

“Log the decision to switch from REST to tRPC for internal services”

remember
topic: auth-architecture
type: decision
content: “Switched from REST to tRPC for internal services. Type safety across the boundary, fewer integration bugs.”
savedid: 2087

“What's on my plate this week?”

list-commitments --status=open

1. Review tRPC migration PR · due: tomorrow

2. Update API docs for v2 endpoints · due: Friday

3. Fix rate-limiter bypass on /auth · overdue by 2d

3ngram MCP·Claude Code·streamable HTTP
Search

Enterprise search for individuals.

Semantic recall across memories and documents. Results ranked by relevance, not recency. GitHub repos, markdown, and structured memories - all searchable through natural language inside your AI chat.

3ngram MCP · project: api-platform
connected

“What have we decided about auth?”

recall “auth architecture”
0.94
Switched from REST to tRPC for internal services
decision · 2d ago
0.91
Auth Architecture Overview
📄 document · docs/architecture/README.md
0.87
OAuth 2.0 with Streamable HTTP for MCP transport
context · 5d ago
0.81
Fail-open revocation mode for staging environments
pattern · 1w ago
4 results·memories + documents·23ms
Personal

Your life, searchable too.

Not just code and work. Travel plans, personal notes, health context, recipes - all searchable with the same semantic engine. Work and personal scopes keep things separate but always accessible.

3ngram MCP · personal
connected

“What was that great restaurant in Florence?”

recall “restaurant Florence”
0.92
Trattoria Mario: “best bistecca we've ever had, no reservations, cash only, get there by 11:45”
highlight · Florence trip · 3mo ago
0.88
Florence & Tuscany Travel Plan
📄 document · travel/italy-2025.md
0.79
“Book the same Airbnb near San Lorenzo market”
note · Florence trip · 3mo ago
3 results·personal memory·19ms

Proactive follow-through

Commitments with deadlines, decisions with rationale, blockers that resolve, and digests that surface what needs attention. Not passive storage.

Cross-provider continuity

Connect once and keep context available across Claude, ChatGPT, Cursor, and every MCP-compatible client. One memory, everywhere.

Search across your work

GitHub repos, markdown docs, and structured memories searchable through natural language from inside your AI workflow. No new UI.

Why 3ngram

Not another memory layer.

Native AI memory handles preferences. 3ngram adds structured follow-through.

Remembers decisions, not just facts

Decisions, commitments, blockers, preferences, patterns. Each with its own lifecycle.

Accountability, not just recall

Tracks due dates, flags overdue items, surfaces stale commitments.

Repeatable briefings and debriefs

Parameterized prompts stored and run via MCP from any client.

Cross-platform via MCP

One memory layer across Claude, ChatGPT, Cursor. No lock-in.

Searches your actual project files

GitHub repos, markdown files, external sources searchable alongside memories.

Product

Your command center

A web dashboard that surfaces what matters, so you spend less time searching and more time shipping.

3ngram dashboard showing focus items ranked by urgency, nudges for upcoming deadlines, and a triage section with overdue and stale commitments
3ngram semantic search showing an AI-generated answer with cited sources and memory results ranked by similarity
Community

Trusted by knowledge workers

I connect 3ngram to Claude Code and every session picks up where the last one left off. No more re-explaining architecture decisions or hunting for context in old threads.

Maren Halvorsen

Senior Engineer, Arcwise

The commitment tracking loop changed how our team ships. Commitments surface in the morning briefing, and nothing quietly drops off the radar anymore.

James Okafor

Technical Lead, Fathom Analytics

I juggle five client projects. 3ngram scopes each one separately, so when I recall context for Acme, I only get Acme. The MCP integration means it just works inside Cursor and Claude.

Sofia Reyes

Independent Consultant

Pricing

Simple pricing

Start free on the Starter plan and upgrade to Individual at $19/month when you need integrations and daily follow-through.

Starter
Free

Full MCP server and web dashboard. No credit card required.

  • 100 memories
  • MCP server (37 tools)
  • Basic search
  • Weekly digest
  • Web dashboard
Get started free

No credit card required.

Individual
$19/mo

For consultants, founders, operators, and solo knowledge workers who need integrations and digests.

  • Everything in Starter
  • Unlimited memories
  • AI answers (RAG search)
  • Integrations (3 sources included)
  • Daily and weekly digests
  • Tags and priority support
Start trial

Founding members get $9/mo for life.

FAQ

Give your AI a Chief of Staff.

Stop losing commitments between sessions.

We use analytics to improve the product. Cookie Policy