Do I need a product management tool?

Maybe not yet. It depends on four sizes: how much feedback comes in, how big your roadmap is, how many people decide and how many agents build. While all four are small, your memory and a tracker do the job, even if the tracker is a markdown file. When one grows past what you can hold in your head, you need one place for the evidence, why it is next and who was told.

TL;DR

• Check four sizes: feedback, roadmap, team and agents.
• If all four are small, stay with what you have. It works.
• When one outgrows your memory, add one place for the why, the evidence and the order, on top of the tracker you already use, whether that is an issue tracker or a markdown file.

What do most small teams already have?

A typical setup for a founder or a small team building with agents:

• Feedback in an inbox, Slack, a support chat and call notes.
• Priorities in an issue tracker, GitHub Issues or a markdown file your agents read, like a TODO.md.
• Building in Claude Code, Cursor, Codex or Grok, with an AGENTS.md that holds your rules.
• Closing the loop with an occasional changelog post, or not at all.

That is a real system. It is cheap, it is fast, and each piece does its job. The tracker knows the task. The rules file knows how you work. What nothing holds is the link between them: the customer behind the work, in what words, why this is next and whether they were told.

Which four sizes decide it?

One size past its limit is a warning. Two or more means the setup is costing you, even if nothing has broken yet.

What are the signs you've outgrown your setup?

1. You cannot say who wanted it. A customer asks "did you do that thing?" and you search Slack, then your inbox, then guess.
2. You shipped something and nobody was told. The person who asked found out weeks later, or never. A changelog post reaches everyone and nobody in particular.
3. Agents wait on you. An agent reaches a fork, which of two behaviors or which customer first, and stops or guesses. The answer is in your head, and you are on a call.
4. You re-explain the same context. Every new session starts with the same paragraph about the customer and what matters. Your rules file holds how you work, not this week's evidence.
5. The loudest customer wins. Whoever wrote last sets the order, and quiet accounts with more revenue behind them wait.

See feature request triage for handling the pile before it needs a tool.

Is it a tracker problem or a deciding problem?

They are different jobs, and mixing them up is how teams buy the wrong thing.

If your trouble is "who is doing this?", a tracker fixes it. If it is "why this and not that?", you have a deciding problem.

What changes with Annsa?

Annsa is the source of direction for your agents. It sits on top of your issue tracker and does not replace it.

• Feedback arrives in one place. Annsa reads Slack, Intercom, call transcripts, surveys, CSV and Google Sheets imports and GitHub, and you can type a request in or send it through an API. Each request keeps the customer behind it, their exact words, the account and its revenue.
• Matching requests group, and the list is ranked. Inside your bets, in your order: what a person marked broken comes first, then what customers asked for, then ideas. You choose; Annsa never decides for you.
• Agents read the same list. Claude Code, Cursor, Codex and Grok connect over MCP, with nothing to install. They claim work, read the spec (the why, the evidence, the Done When lines) and record evidence against each line.
• Shipped work reaches the person who asked. When a job ships, Share Back drafts a note to each person behind it. You review and send.

Your tracker stays. GitHub issues and pull requests come in as signals.

A "have I outgrown my setup?" checklist

Two minutes. One question per size, plus one on telling people. Answer yes or no, then count.

Have I outgrown my setup? (yes or no)

Feedback   Requests reach me in three or more places,
           or I don't read them all.                         [ ] yes  [ ] no
Roadmap    I can't say who wanted each of my next
           ten jobs, or why one sits above another.          [ ] yes  [ ] no
Team       Someone other than me decides what gets built,
           or talks to customers without me.                 [ ] yes  [ ] no
Agents     Two or more agents build at once, or an agent
           waited on a decision from me for a day.           [ ] yes  [ ] no
Told       I shipped something a customer asked for in the
           last 30 days and didn't tell that customer.       [ ] yes  [ ] no

Count of yes: [ ]
0-1  Stay put. Keep writing down exact words and names.
2-3  Fix the one that hurts most first.
4-5  Put the evidence, why it's next and who was told in one place.

What should you do this week?

1. Run the checklist above.
2. If you stay put, start writing down exact words and names beside each request. It takes a minute and keeps the option open.
3. If you move, start with one thing: tell the people who asked for the last thing you shipped. The closing the feedback loop audit takes ten minutes.
4. When one place for the evidence, why and who was told would help, Try Annsa. The free trial covers 100 customer signals, once.