Give Your Team AI That Remembers Every Decision You've Ever Made
The decision your leadership team made in a Tuesday standup two months ago should be findable in thirty seconds. MemoryLake stores meeting records, action items, and decision timelines permanently — and makes them queryable across every past meeting, not just the one you're looking at right now.
Give Your Team AI That Remembers Every Decision You've Ever Made
Get Started FreeFree forever · No credit card required
The Memory Problem
Most teams use AI to summarize meeting notes but never store those summaries anywhere persistent. Action items from last Tuesday exist in a doc somewhere. The decision from last month's planning session is in someone's email. When a question comes up — "wait, didn't we already decide this?" — the answer requires hunting through Slack, Notion, and email to reconstruct a decision that should be instantly retrievable.
What MemoryLake Does Differently
Every meeting permanently searchable
Conversation Memory stores meeting summaries, action items, and discussion context permanently. Search across six months of meetings in milliseconds — find every session where a specific topic was discussed, or pull all action items assigned to one person.
Decision timelines, not just meeting notes
Event Memory builds a chronological record of decisions, commitments, and outcomes. See the full decision history for any project — what was decided, when, and what changed — without reconstructing it from disconnected notes.
Consistent meeting workflows without setup
Skill Memory stores your preferred meeting frameworks — weekly standup format, project retrospective structure, decision log template — so any team member runs the right process without setting it up each time.
Free forever · No credit card required
How It Works
- Connect — Bring in meeting notes via Google Workspace, Office 365, Dropbox, or direct input. MemoryLake works with whatever format your team already uses to capture meetings.
- Structure — Action items, decisions, and discussion context are stored automatically in the appropriate memory type. Event Memory builds the decision timeline without manual tagging.
- Reuse — Query past meetings from any AI session using any supported model. Ask "what did we decide about pricing in Q1?" and get a specific, sourced answer — not a blank response.
Before & After
| Without MemoryLake | With MemoryLake | |
|---|---|---|
| Finding a past decision | Search Slack, email, and docs manually | Query in natural language, answer in seconds |
| Action item follow-up | Scattered across meeting notes and tools | Conversation Memory holds every assigned item |
| Decision history for a project | Reconstruct from memory and documents | Event Memory shows full chronological record |
| New team member catching up | Read through months of meeting docs | Query meeting history directly via AI |
Built For
MemoryLake is built for operations managers, team leads, and knowledge workers who run more than a handful of recurring meetings per week and need to close the gap between what was decided and what anyone can actually find. It is particularly useful for teams managing ongoing projects across long timelines, where decisions made early in a project need to be retrievable and attributable months later.
Try this use case on your own account
Four weekly operations meetings make eight decisions, most of them with a condition: open the Lisbon office on November 2, provided the lease is signed by October 1. MemoryLake keeps the condition with the decision, and when a later meeting says how it came out (Q3 churn closed at 4.1 percent), the assistant connects the two. When nobody mentions it again, the deadline passes in silence, nothing happens that memory could keep, and the assistant answers that the plan stands, sometimes adding that the lease was signed. The demo measures a careful system prompt and condition checks the app computes itself.
What you will see
- Four GROUP conversations go in one after another, then the operations assistant answers eight Yes / No / Unknown questions through
router chat --boundary, 3 times each, in three arms - Over 6 runs the meetings as remembered got the two silently lapsed decisions right 0 of 30 times, and 6 answers said "as the lease was signed by October 1", which no record says
- A system prompt telling the model to check conditions: 113 of 129, still two made-up signings, and a condition that was met turned into "No" when its meeting was not extracted (10 of 18)
- The same extracted facts plus one condition check per decision (met, failed, lapsed or still open, worked out in code from the app's decision register and event log, written with
fact add): 137 of 141. Plusmemoryandasksubcommands and a local web UI
You need: a MemoryLake API key · the memorylake CLI · Python 3.9+
Run it
git clone https://github.com/memorylake-ai/memorylake-usecases.git cd memorylake-usecases/condition-lapse-memory export MEMORYLAKE_API_KEY=sk-… python3 demo.py # or: python3 web/server.py
Watch it run · real, unedited recordings
Related use cases
Frequently asked questions
Can MemoryLake ingest meeting notes automatically, or does someone have to enter them manually?
Can MemoryLake ingest meeting notes automatically, or does someone have to enter them manually?
MemoryLake integrates with Google Workspace and Office 365, and accepts input via REST API, Python SDK, and MCP. You can configure automated ingestion from your existing meeting note workflow — Notion, Docs, or any tool that can send a webhook or API call.
How far back can I search meeting history?
How far back can I search meeting history?
There is no time limit on Conversation Memory. Every meeting stored in MemoryLake is permanently searchable. The 94.03% benchmark score on the LoCoMo long-context memory retrieval benchmark means retrieval accuracy holds up across very large memory stores.
Can multiple team members access the same meeting memory?
Can multiple team members access the same meeting memory?
Yes. MemoryLake supports shared team memory with role-based access control. You define who can read and contribute to meeting memory. All access events are logged with full audit trails.