“Didn’t we already decide this?”
The shared decision recordfor your team and its AI agents.
You’ve already thought this through. Your next agent should know why.
“Build a separate version for this customer.”
Keep one shared product.
So fixes reach every customer.
Add a customer setting. They still get the same fixes.
A separate version means revisiting the decision.
Give it your team’s recorded decisions, trade-offs and what changed.
Keep working in your connected agent.
Early beta is currently free. Pricing is coming.
For teams building and working with agents.
When the work changes hands
Your team keeps explaining the same decisions.
A feature moves through Design, Engineering and customer conversations. The choice, reason and scope are still scattered across people and tools.
“What did we agree?”
Design picks up the feature.“Why this approach?”
An agent starts the implementation.“Who can change it?”
The customer team gets a new request.
The next person—or agent—has to put the backstory together again.
We build Product Brain with Product Brain
A recorded choice.
Reviewed code.
We changed how much context a task briefing could carry. The recorded decision guided the implementation; independent review found a retry mistake, which was corrected before merge.
The briefing needs
more room.
- Agreed
Keep the essential context.
A task briefing needed more room, while its direction and rules stayed bounded.
- Built
Change the briefing behavior.
The implementation followed that choice and its limits.
- Reviewed and corrected
Fix the retry advice.
Review caught advice that sent an agent back to a limit it had already reached. We fixed it and added two tests.
Our own development workflow. We’re still testing its effect on quality and rework.
See the development trail
Accepted decisions DEC-2124 and DEC-2125 explain the change to the default task briefing. TASK-916 scopes the implementation. Product Brain pull request 1832 implements it and records the decisions it follows.
Independent code review found retry advice pointing back to an exhausted tier. The implementation was corrected and two tests were added before merge. This is evidence of a recorded reason shaping reviewed development work.
How a recorded decision helped write this page
A fresh agent looked up our positioning and drafted an earlier introduction for this page.
Our own writing task, not a customer result. Read-only lookup · 11 October 2026.
The agent ran pb orient --task 'Write a homepage intro for Product Brain. Who is it for, what does it do, and why should they use it?' --full --json, then opened the relevant records in full with pb get-many. CLI 0.1.0-beta.9860.
Selected unchanged text from DEC-1924:
“agent-first teams are teams where every person works through agents, across proposals, contracts, operations, program design and marketing analysis, not only engineering.”
The agent read the existing decision, its owner, related definitions and history. It followed up where the first briefing was incomplete.
The headline was already approved. We asked for the supporting paragraph and edited the draft. The lookup left the record unchanged.
Early use · UpCurrent
Their first impressions.
On governance and decision-making
“This is governance heaven.”
On using Product Brain
“Product Brain gives me so much peace of mind”
Knowledge stays with the team
They’re away.
The reason isn’t.
The person who made the choice can be unavailable. Their recorded reason stays available to the team and its connected agents.
See shared reuseOne-off export first.
The reason stays with the choice.
Build within scope
Explain what’s included
Why use Product Brain alongside your tools?
Which version are we building from?
Your brief, code and team update can drift apart. Keep the current choice, reason and owner together, so people and connected agents can check the same agreement wherever they do the work.
One-off export first.
Learn before adding scheduling.
- Product brief
One-off export first.
- Implementation
Scheduling stays out.
- Leadership update
First release: one-off export.
What about the tools we already use?
Your documents, repository instructions and project tools can already hold decisions and reasons. Keep the detailed work there. Product Brain gives the choices you reuse across them an owning role, change history and access through your connected agents. Your team still keeps that record current.

Why this exists
The decisions were there.
Keeping them useful was a job.
I saw a connected record produce useful briefs and surface contradictions. Keeping it current became a job. I built Product Brain so your agents can help keep that record useful while you work—with clear authority for changes.
Talk with RandyConnect your agent.
Keep working.
Start with one real decision, in a task you already need to do.

