The latest decision wins.
Jackdaw resolves contradictions and knows when a newer decision supersedes an older one—without erasing the history.
Jackdaw keeps the important facts about you and your work current, sourced, and under your control—so people and agents can act without guessing.
We save what you tell us. We ask about what we figure out.
Your inbox has the old plan. Slack has the correction. The meeting changed it again. Finding a matching fact is not the same as knowing what is current, approved, and safe to use.
See how the Trust Engine resolves truth →Your tools already store the history. Jackdaw resolves it into one current, sourced, approved truth—then gives each person and agent only what their task and permissions allow.
Jackdaw resolves contradictions and knows when a newer decision supersedes an older one—without erasing the history.
Facts remain connected to the messages, meetings, documents, and people that support them. Uncertainty stays visible.
Agent proposals wait for review. Each person and agent receives only the context the task and its permissions allow.
The living wiki gives people a place to inspect and shape the truth. Jackdaw then carries the current decisions, owners, and constraints into every connected agent.
Today, 10:00 AM · Prepared from the Q3 launch, Checkout v2, and Northwind feedback. [14 sources]
Guest checkout: out of scope.
Ship Aug 1. Maya owns the invite-only rollout.
Drafted the recap with the Aug 1 date, the invite-only rollout, and the decisions from the room.
Updated the build plan for guest checkout and kept every change behind the approved flag.
Prepared the Northwind brief from what changed, why it changed, and the source behind each claim.
One page · current everywhere · proposed changes come back for approval
Use Jackdaw for one—or all three. In every space, people and agents work from the same current, trusted context.
Your plans
Your household
Your team

Not the name and the birthday—those are easy. It’s the last DTaP date buried in a PDF, the district you moved from, and the box that wants a paragraph about your kid. Jackdaw has all of it, because you already said it somewhere.
The same for job applications, medical intake, insurance claims, and the thirty change-of-address forms nobody finishes.
Every deal, the same security review in a slightly different spreadsheet. The answers are scattered across Slack, Notion and a SOC 2 PDF — Jackdaw fills the Response column from what your team actually decided, and puts the receipt in the next one over.
The same for RFPs, vendor onboarding packets, compliance evidence requests, and the insurance claim you have to assemble in a bad week.
| Question | Response | Evidence | |
|---|---|---|---|
| 14 | Do you enforce SSO for all vendor access? | Yes — required of all vendors since January 2026. | #security · Jan 8 |
| 15 | Customer data retention period? | 90 days, then hard-deleted. No soft delete. | Notion · Data policy |
| 16 | Most recent third-party audit? | SOC 2 Type II, March 2026. Report attached. | Drive · SOC2-2026.pdf |
| 17 | Encryption at rest? | AES-256, keys rotated quarterly. | Notion · Infra |
| 18 | Do you maintain a sub-processor list? | we have never decided this — ask Marcus | — |
| 19 | Breach notification window? | 72 hours to the named contact. | MSA §11.4 |
It lives in commits, tickets, a churn thread and a call nobody wrote up. Jackdaw drafts it from what actually happened, with a link behind every claim — and tells you the one thing nobody recorded, instead of writing around it.
The same for board and client reports, performance reviews for you and every report, postmortems, and the hours you forgot to bill.
Shipped. Audit logs landed the 14th — the last blocker on three enterprise deals. [Linear · ENG-441]
Lost. Westbrook churned on pricing. Second this quarter, both mid-market, both on the same objection. Worth a call before it’s a pattern. [#deals · Jun 21]
Team. Two engineers started the 2nd, both on the platform side. Ramp has been faster than the last two hires. [Gmail · offers]
I couldn’t find why pricing changed in May — it was decided on a call and nobody wrote it down. You’ll want a line here; it’s the question they will ask.
When the conversation ends, the agent changes, or the plan moves on, Jackdaw carries the current decisions, preferences, owners, and constraints forward. The next person or agent picks up where the last one left off.

Support keeps the same policies. New teammates see the decisions behind the work. Change something once, and every connected tool catches up.
Add retry + idempotency key to webhook handler
priya-raman merged 3 commits into main

Training plans adjust to the week you actually had. Projects remember the decisions and hard no’s. Routine work moves without another round of setup.
4 bd · south-facing yard · 12 min to school
Disclosure p.14 — original knob-and-tube wiring
“Sorry,” Elena said, and did not look at him.
Her grey eyes caught the light off the café window, and for a moment she was somewhere else entirely.

Health details and care notes stay easy to find. Pickups and schedule conflicts surface before the group chat turns into an emergency.
Milo’s allergic to penicillin. Ivy’s inhaler is in the blue bag — she carries it. Beds 8 and 8:30. Dr. Nguyen (209) 555-0148.
Same engine, same privacy, same cross-provider model — for you, your household, or your whole team.
For solo work across the AI tools you connect.
Up to 5 members. One shared memory for the household.
For teams that share context across the work.
Closed beta pricing. Locked in for early members.
Provider memory is useful, but it stays inside one provider and mostly learns from what happens there. Jackdaw keeps approved facts, constraints, corrections, and rejections in a visible wiki above your AI tools. Connect another compatible client and the same durable context can come with you.
Yes. Jackdaw exposes your model via standard context protocols (MCP and provider-native APIs). If your tool can take a system prompt or call an MCP server, it can use Jackdaw.
Corrections stay in the current conversation first. Approve one for your long-term wiki and every connected provider can inherit the corrected fact instead of the value it replaced.
Never. Your model is yours alone. We do not use your personal data to train any foundation model, improve our product for other users, or sell to third parties.
Yes, completely. One request purges your structured memory, vector index, and message history — not a soft delete, a full cascade across every store. You can also disconnect individual sources at any time.
All connections are encrypted in transit via TLS. Disks are encrypted at rest. The architecture enforces hard per-user data isolation at every database layer — there is no code path that can return one user's data to another.
Jackdaw is in closed beta. Request access and we’ll reach out personally when a spot opens.