Private beta — founders first, scales with the team

The memory of why.

You decided something six months ago. The decision survived; the reasoning didn't. Familiar reconstructs it from GitHub, Slack and your own scribbles — what was chosen, what was rejected, and why — and answers with citations.

Limited to 20 engineering teams

The problem

The decision survives. The reasoning doesn't.

Made in a meeting, half-noted in a thread, shipped in a PR — and by next quarter nobody can reconstruct why.

Three months later — “wait, why did we choose Postgres?”
And you are the only one who knows. Until you don’t.
25% of the workweek goes to searching for information.
56% say the only way to get context is to ask a person.
1 in 2 teams unknowingly redo work already done.
Atlassian, State of Teams 2025 · 12,000 knowledge workers · 2.4 billion hours a year across the Fortune 500
The product

Ask. Get the answer. See the receipts.

Every answer names the alternatives that lost, and cites where it came from.

familiar — decision memory
Try a question

Interactive preview with sample data.

Why it's different

Search finds documents. We reconstruct reasoning.

The rejected alternative is almost never written down — so there's no passage to retrieve. It has to be reconstructed from evidence, then confirmed by a human.

What we store
A decision — with the options that were rejected, the reason each lost, who made the call, and what happened after. Not a passage that mentions it.
Human confirmed
Nothing enters permanent memory without a person approving it. That's what makes the graph trustworthy instead of just large.
Private by default
Your own memory stays yours. A decision reaches the team only when you promote it — one deliberate click.
Vendor neutral
Microsoft will win inside GitHub. Atlassian will win inside Confluence. Neither serves a team living across GitHub, Slack and Notion — which is most teams.
Why now

AI made code cheap. It made context scarce.

Every year more of your team's reasoning happens inside a chat window that forgets it. The code ships. The why doesn't.

Nothing survives the session
"AI conversations are ephemeral by design — decisions made early fade as sessions lengthen, and nothing survives the session boundary."
Thoughtworks / martinfowler.com, 2026
More code, less rationale
AI-co-authored pull requests average 1.7× more issues than human-only ones — and arrive with no record of what was considered or rejected.
CodeRabbit, 2025 · 470 real-world PRs
Speed without context backfires
In a randomised trial, experienced developers were 19% slower with AI tools — and believed they had been 20% faster.
METR, 2025 · 16 developers, 246 tasks
Almost right, but not quite
66% of developers say their top AI frustration is code that is almost right; only a third trust its accuracy.
Stack Overflow Developer Survey 2025 · n=49,009
Two memory spaces

Start solo. Scale with the team.

Founders first

Free, and it stays free. Capture a half-formed thought at 11pm — typed or spoken. Weeks later, when the PR lands, Familiar links the seed to the decision it became. Ask what you decided this quarter, and why, with sources.

Then the team

Past roughly fifteen people the team layer becomes the product. Promote a decision and it becomes institutional memory — cited back to the PR and thread it came from. New hires stop asking you. Settled debates stay settled.

It comes to you

The digest doesn't summarise. It notices.

Anyone can list what you did. Because we store decisions with their alternatives and outcomes, we can show you the patterns underneath them.

Early access

Join the first teams using Familiar.

We're onboarding a small group of engineering teams as founding design partners. You'll shape what gets built.

Founding design partner
  • Limited to 20 engineering teams
  • Direct line to the founders, weekly
  • Free through the beta, then partner pricing
  • Your workflow shapes the roadmap
Founder Engineer Eng leader Other

No spam. Only email about early access.

You're in.

We'll be in touch as partner slots open — usually within a week.

Questions

GitHub and Slack at launch, plus direct capture — typed notes and voice. Meetings, Gmail and AI-chat history import come after. We'd rather do two sources well than six badly.

You control every source you connect, and permissions are checked before an answer is generated, not after. Personal memory and team memory stay separate — only what you promote crosses between them. Nothing becomes permanent memory without a person confirming it.

No. Search finds documents; Familiar reconstructs reasoning. It links a decision to the alternatives that were rejected, the constraints that shaped it, who made the call, and what happened afterward — then cites all of it.

It's the hardest part of the product and we're honest about it: extracting a decision and its rejected alternatives from a messy PR thread is an inference problem, not a lookup. That's why a human confirms before anything becomes permanent memory. Design partners see the raw evidence behind every proposed decision.

Free for founding design partners through the beta, then partner pricing at launch. We'll agree pricing with you before anything changes.

Kevin Jojo and Ishan Chaskar. We met as computer-science classmates in 2021 and have both shipped AI products end to end. Email us directly at kevin@familiar.ai.

The vision

Every AI starts from zero.
Familiar is the memory that means it never has to — first what you decided, eventually how you think.