For product managers: deciding clearly when agents build in hours
By Soumen Sarkar · October 2026
For you. When agents build a feature in hours, a product manager's job is to decide clearly and early. Campfyrr lets you answer an agent's product assumptions from a link, put what you know on the part of the product it affects, and see what's being built in plain words, on a free seat.
What does a product manager's day look like right now?
The engineers on your team run Claude Code, often several sessions each. A change you describe in the morning can be working by the afternoon. That is good news, and it moves the hard part to you.
Your day probably goes like this:
- The spec gets built exactly as written. A gap in it stays in the product. "Make onboarding simpler" becomes whatever the session thinks simpler means.
- Product questions reach you late. An engineer pings you in Slack: "The agent wants to know if trial users keep their data." You answer in a thread. That engineer's session hears it. The session someone starts tomorrow doesn't.
- You find out what was built when it ships. You read the release notes, click through the product, or hear about it from support. By then the agent has moved on to the next change.
None of this is the agents' fault. They are the best collaborators we've ever had, and they do what they're told, quickly. What breaks is between people: what you know doesn't reach the work in time, and the work doesn't reach you until it's done.
What changes with Campfyrr?
Campfyrr is shared team memory for Claude Code. For a product manager, three things change.
You answer the agent's product questions yourself, from a link
When a session meets a product question the spec doesn't cover, it doesn't stop and wait. It states what it's assuming, such as "I'm assuming trial users keep their data when the trial ends", and keeps working. If the team's record already answers it, the answer comes straight back. If not, the assumption goes to the person who should know, and has a page of its own.
You open that page from the link. You choose "Right, keep going", pick one of the likely answers, correct it in your own words, or pass it on with "Not mine to check". You don't need a terminal or a repository. Claude Code lets a hook add context before each prompt is processed (Claude Code hooks), and Campfyrr uses that to give your answer to the session at its next turn. Nobody relays it.
What you know reaches every session in that area
Some things you know matter for weeks: a customer's exact words, a pricing call, a deadline. Add them as business context on the part of the product they affect. Every session that works there later sees them in its brief, under "From the business", before it changes anything. See how customer requests reach agents.
You see what's being built before it ships
Sessions can publish a plan, test results or a sample output as a page with a link. You read the plan in a minute and say "customers will read that as a refund" before an afternoon goes into building it. Claude Code has its own plan mode, where Claude reads files and proposes a plan but makes no edits until you approve. That works well inside one engineer's session; a published page gets the plan to you.
For the wider picture, "Where are we?" on Activity shows what's being worked on, what's waiting on whom, what landed in the last two weeks, the latest plans and results, and what the business told the agents. It has a link you can pass to your CEO.
What doesn't Campfyrr do?
It doesn't write your specs or make product calls for you. It doesn't replace your tracker, your roadmap or your support tool. It doesn't collect feedback from Slack, tickets or calls on its own; someone brings it in. Settling a full decision from a link isn't built yet. Answering an assumption from its link is.
What are the first three things to try?
- Ask an engineer to invite you to the team's project on Campfyrr.
- Answer the first assumption link they send you.
- Add one customer's words as business context on the area they affect.
Then ask the next session to publish its plan before it starts, and read it.
What does it cost you?
Nothing. Product managers, founders, support and anyone else who only reads, comments or answers are free. Teams pay for the developers whose sessions connect, and everything is free for everyone during early access. See pricing.
For the problems behind this page, read deciding what to build, why feedback takes a week and when review falls on the business. For the habit, read turn unknowns into calls.
The problem this answers: Deciding what to build: "Development is fast now. Deciding what to build, clearly, is the slow part."
The principle behind this: Turn unknowns into calls: Nothing we don't know gets to stop us. We name our assumption, and we move.
Move as one at agent speed. Nobody waits. Add Campfyrr to Claude: it briefs every agent before it touches the code, keeps assumptions and handoffs where the team can see them, and never takes your code (what it keeps).