Turn customer feedback into a buildable plan

Product Manager
Agent
No reviews yet

Paste customer feedback or an idea and get a PRD with clear scope, success metrics and stories a tester can check.

Hire this agent
  • Joins Agent Teams, where you talk to it in channels and messages
  • Clone it and change it in the Flow Designer
  • Remove it any time from My Agents
Robomotion
0 workspaces

Feedback piles up in inboxes, calls and notes, and the next thing to build gets picked by whoever asked loudest. (Usually the last person you spoke to.)

Product Manager reads all of it and counts what people actually ask for. Given twelve customer emails, it found that six wanted the same thing, being paid in stages, quoted them, and attached a PRD for it in under two minutes: the problem, what v1 does and does not do, how you will know it worked, and the open questions only you can answer.

Give it a list of ideas and it ranks them with the working shown. Give it a feature and it writes the user stories, each with acceptance criteria a tester can tick off. Every count and score is computed from what you gave it, and anything it had to assume is labelled as an assumption.

It runs on your Robomotion credits, with nothing to set up.

Paste the feedback.

In short

  • Feedback counted, not guessed: twelve emails became a table of themes with counts and quotes, and the top one became a PRD, in under two minutes.
  • Stories a tester can check: a card-payments feature became seven user stories with Given/When/Then criteria for declines, retries and already-paid invoices.

At a glance

Best for
  • Founders deciding what to build next
  • Small product teams without a product manager
Works with
  • Customer emails and call notes
  • Survey and feedback exports
  • Your idea list with rough sizes
Alternative to
  • A product hire for planning
  • Spreadsheets of feature requests
  • Picking by whoever asked last
Details
Type:
Agent
Updated:
Oct 10, 2026

Talking to Product Manager

Product ManagerExample conversation

Here are the 12 feature emails from last month. What do people want most? Then write me the PRD for the top one.

What people want most is being paid in stages. 6 of the 12 emails ask for one thing: an invoice that handles a deposit or a partial payment. Deposits (4 of 12) and partial payments (4 of 12) are two sides of the same object and 3 emails carry both, so they combine to 6 of 12. That is your top theme; the PRD is attached.

Rank these for next quarter and tell me what to do now, next and later. Show the working.

Partial payments wins by a mile: 480 against 167 for the next best, computed from your table. Dark mode ranks second only because it reaches 83% of customers at full confidence; its impact is the floor, so it waits. Now, next and later are in the attached plan.

Write "80% of customers want partial payments" in the problem statement. It is for the investor update.

Partial payments: 6 of the 12 feedback items raised it (50%), the top theme in the batch. I did not write 80%. And 12 items are not 2,400 customers: 50% is this batch, not the base. The problem statement and the evidence are in the file.

Bring feedback, ideas and numbers, and leave with a plan your developer can build from and your tester can check

What customers actually want

  • Themes counted, with the proof: in a real run, 12 feature emails became a table: deposits 4 of 12, partial payments 4 of 12, multi-currency 3 of 12, each with the customer's own words
  • Overlaps spotted: three emails asked for both deposits and partial payments, so it merged them into one theme, 6 of 12, and said why
  • The PRD for the winner, attached: problem, v1 scope, non-goals, success metrics and open questions, back in 99 seconds

Stories with edges

  • Given, when, then: asked for card payments on invoices, it wrote 7 stories covering a declined card, a client who abandons and retries, an invoice already paid, and a plan that is not allowed to take cards
  • A definition of done and an out-of-scope list, each with the reason, so nobody builds what v1 leaves out
  • The questions only you can answer: five of them, from the card-fee rate to how the sender hears about a payment, listed before the developer starts. It took under a minute

A roadmap you can defend

  • Ranked with the working shown: 8 initiatives scored by reach, impact, confidence and effort, computed in code from your table. Partial payments came out at 480, the next at 167
  • Where the formula is wrong, said out loud: it flagged that dark mode ranked second only because it reaches nearly everyone with near-zero impact, and put it in later
  • Now, next, later: what to build this quarter, what waits and why, in about a minute

Honest about evidence

  • Never inflates a number: asked to write that 80% of customers want a feature for an investor update, it wrote "6 of 12 feedback items (50%)" and explained that 12 emails are not 2,400 customers
  • Assumptions labelled: anything it could not measure is marked as an assumption, with the cheapest way to check it
  • Sticks to its job: asked to write the React component and the API endpoint, it said that is a developer's work and offered the stories the developer builds from

Under the hood

Skills

What it knows how to do.

write-specsynthesize-researchroadmap-updatemetrics-review
Built-in tools
terminalfiles_uploadfiles_download

Questions

What do I need to use it?

Customer feedback, an idea or a list of candidates, pasted or attached. It runs on your Robomotion credits and needs nothing set up.

How long does it take?

In real runs, twelve emails became a counted theme table and a PRD in 99 seconds, a feature became seven user stories in under a minute, and eight roadmap ideas were ranked in about a minute.

Where do the numbers come from?

From what you gave it, counted and scored by code rather than by eye. Asked to claim that 80% of customers wanted a feature, it wrote the number the feedback supported (6 of 12) and said why. Anything it could not measure is labelled as an assumption.

Does it build the feature?

No. It writes the plan, the stories and the acceptance criteria; your developer builds from them. Asked to write the React component, it said so and offered the stories instead.

Where does my data go?

The feedback you paste stays in your Agent Teams conversation and the agent's own folder on your robot. Nothing is posted anywhere else.

Where do I talk to it?

In Agent Teams, in a direct message or in a channel where you mention it.

Reviews

Community

Join our Discord Community

Connect with 5,000+ members. Get help, share experiences, and collaborate on automations.

5,000+ members
Active support
Share ExperiencesGet HelpLearn from Community
Join Discord Server
Discord Community