One page your team keeps — and every agent answers from.
Projects, decisions, people, and the evidence behind them. Your team edits it directly, every agent reads it, and nothing lands on it without someone approving first.
A living wiki your team and your agents share.
Open one page and know where the work stands.
See the owner, current state, decisions, guardrails, next step, sources, and edit history together. Your team can improve the page directly, and connected agents work from that same approved context.
What is true now—not last quarter’s summary.
The call, the reason, and where it came from.
Who owns the work, what changed, and what happens next.
One-page checkout, replacing the old three-step flow. Owned by Payments — DRI is @Maya. Ships behind a flag for the Q3 launch. [Linear]
- —Use Stripe’s prebuilt checkout — not a custom form. [Slack]
- —Guest checkout: out of scope.
- —Tax handled by Stripe, not in-house.
Ship behind the flag by Sep 1.
- —Never store raw card numbers — Stripe holds them.
- —Every change ships behind the flag. [Notion]
Claude Code · building checkoutPulled Decisions + Guardrails — builds on Stripe checkout, now includes guest checkout.
Cursor · PR #214 reviewPulled Guardrails — flagged code logging a full card number before anyone reviewed it.
ChatGPT · release notesPulled Now — the draft says “ships Aug 1,” the date Maya edited seconds ago.
A team should not need another reporting ritual to reconstruct what it already knows. The page becomes the handoff: managers see the current state, teammates improve it, and connected agents read the approved truth.
Notes should become a network, not a pile.
Conventional notes are documents you file away. Wiki-style notes turn every page into a node: a project links to its people, a decision links to its source, and each page shows everywhere it is referenced. The knowledge compounds as you write.
Jackdaw adds what an ordinary wiki cannot: connected evidence around the page, reviewable suggestions from conversations and external AIs, and the same approved network available to compatible agents.
A place to think, write, and connect ideas.
Headings, checklists, tables, code, callouts, slash commands, mentions, and rich formatting make it a serious place to think and write. Underneath that editor, [[links]] and backlinks turn the notes into a navigable model of your work.
A three-week push around the launch. Goal: 1,200 new signups, cost per signup under $20.
Slash commands, [[links]], and @mentions make the wiki useful to people. Those same pages become memory your agents can read.
The system can suggest. You decide what becomes durable.
Conversation-caught facts and proposals from external AIs land in your Inbox. Approve or reject each one before it joins your long-term wiki. Direct edits still save immediately—because you made them.
Review captured facts
Approve or reject conversation captures and proposals from connected AI tools.
Apply proposed page edits
Assistant edits show Apply or Discard before the page changes.
Restore page history
See who changed a page, when it changed, and restore an earlier revision.
API layer uses tRPC exclusively. No REST endpoints.
Renewal owner: Maya Chen → Raj Patel
Trust what it knows.
A shared brain only works if you believe what it says. So every fact carries its source, every edit can be undone, personal context stays sealed off from the team’s — and everything is encrypted on the wire and on disk.
“Confirmed with Raj — they’re renewing in Q3 and bumping to the $240k tier.”
Inferred facts keep the source that supports them.
TLS 1.3 in transit, AES-256 at rest, and hard per-user isolation at every database layer.
Your data is yours. We don’t train models on it, and we never share it.
One request pulls everything out — or purges it completely. Anytime.
Your team writes the context your agents use.
One wiki. Two kinds of collaborators.
Build an agent in Jackdaw or connect one over MCP. Both work from the same approved pages your team can read and edit.
Agents start from approved pages.
Built-in agents and compatible MCP clients retrieve the decisions, constraints, and history relevant to their work.
Durable changes come back for review.
When an agent learns something worth keeping, it can stage a proposal in your Inbox instead of silently rewriting the wiki.
See the rest of Jackdaw.
Give your agents a memory worth keeping.
Jackdaw is in closed beta. Request access and we’ll reach out personally when a spot opens.