Issue · 02 · Spring Knovya · Letters
Letter · 03

AI for product managers who keep score.

— Spring, the year of forgotten decisions

Built for the PM with twelve open tabs, a backlog they half-remember defending, and a Slack thread asking "why did we choose Postgres over DynamoDB again?" One letter, in three parts: what we noticed, what we built, and what your next sprint gets back.

The Letter

Dear Product Manager, on AI for PMs and decisions that stick

Friday afternoon. Engineering messages: "hey, why did we choose Postgres over DynamoDB again?" You know you decided this. You were in the meeting. You wrote a doc. The doc is in Notion, or maybe Google Drive, or maybe the comments of a Linear ticket from October. You dig for twenty minutes and find three contradictory versions. The decision evaporated.

Not because you didn't write it down — you did. Because it lived in twelve places, and the version that lived in your head didn't match what shipped. Six months later, you're rebuilding context that was never lost — just scattered. The AI tools for product managers on the market layer more chat on top of more documents. They write the next PRD. They don't remember the last one.

And the decision in October wasn't just a database pick. It was the customer call from week eleven, the trade-off note in the retro, the architecture sketch on a whiteboard, and the Slack thread where the staff engineer pushed back. That's the decision — not the line that says "we chose Postgres." When any one of those pieces moves, the decision moves with it. You need the connective tissue, not another empty document.

Knovya makes decisions stick. Decision Log templates catch every choice with the option you picked, the alternatives, the reasoning, the trade-off, the revisit threshold. Experience Envelope groups past decisions by outcome, so when the same question arrives in a new shape, the precedent surfaces itself. NoteRank ranks the decision archive by relevance — the one you need surfaces before you finish typing the question.

And the rest of the workflow comes along. AI Co-Edit drafts the PRD with full context — past decisions, related specs, the customer voice from interviews you transcribed last month. Conversation→Note turns a Slack thread or a Claude session into a structured decision card, automatically. The knowledge graph connects the decision to the spec, the meeting note, the ticket — all of them findable from any one of them.

Stop rebuilding context. The decision is already written — Knovya just keeps it findable.

— Knovya

The Decision · Try it

A decision, six months later.

This is what surfaces when engineering asks the question again. Switch the lens to see how the same decision lives across views.

decision · KNOV-241
— surfaced from your archive
KNOV-241 DECISION · OCT 14
SHIPPED

Use Postgres over DynamoDB for the orders service

Context
Need ACID, joins across line items, complex analytics queries on order history.
Picked
Postgres on RDS, multi-AZ, read replicas.
Considered
DynamoDB (faster scaling, weaker query model), CockroachDB (strong but operational cost).
Trade-off
Slower horizontal scaling. Cheaper at our size. Engineering team already fluent.
Revisit
If sustained write rate > 5k/sec, or if we ship multi-region in 2026.
3 docs linked 2 retros referenced 5 connected notes

Three lenses, one decision. The card is what you wrote. The envelope is what the system remembered. The thread is what becomes a decision tomorrow.

The Stack — six things, one PM workflow

From the Slack thread to the PRD, in one workspace.

Decision capture, retro recall, customer voice, roadmap precedent, and the AI that connects them. Built for the work between meetings.

  1. 01

    Decision Log

    Templates that catch every decision with the option you picked, alternatives, trade-offs, and a revisit threshold. Each card links to the spec, the meeting, the retro, the ticket. The decision lives in one place — and finds you when it matters.

    Decision log →
  2. 02

    Experience Envelope

    Past decisions grouped by outcome — what shipped, what reverted, what's still open. When a question arrives in a new shape, the envelope surfaces the precedent. The retro happens before the meeting, not after.

    Experience envelope →
  3. 03

    NoteRank

    Ten signals rank your archive — graph density, recency, your own engagement, and what you marked as a precedent. The retro you forgot you wrote surfaces before you finish typing the question.

    NoteRank →
  4. 04

    AI Co-Edit

    A side panel that drafts the PRD with full context — past decisions, related specs, customer interview voice, the roadmap where this fits. It writes with what you already wrote, not from a blank prompt.

    AI Co-Edit →
  5. 05

    Conversation → Note

    A Slack thread, a Claude session, a customer interview transcript — converted into a structured decision card or research note, automatically. Capture stops being a chore. The thread becomes a searchable artifact.

    Conversation→Note →
  6. 06

    Knowledge Graph

    The decision links to the spec links to the retro links to the customer voice. Any one entry is the door to all of them. The PM stops being the connective tissue — the graph is.

    Knowledge graph →
A Week, in Practice

Sprint 14, somewhere between roadmap and retro.

You ship every two weeks. Fourteen sprints in, the decisions are compounding — the ones you keep, and the ones you forgot you made. Here's the week, in seven scenes.

  1. Mon · 09:30

    Standup

    Engineering flags a blocker on the orders service. You log it as a decision-needed card — context, who's affected, the shape of the question. The card auto-links the original KNOV-241 from October.

  2. Tue · 11:00

    Customer call

    A founding customer needs multi-region. You record the call and let voice transcription run. The note lands with named quotes. AI Co-Edit summarizes themes against the last six interviews — the pattern is consistent.

  3. Tue · 16:30

    Roadmap call

    Leadership asks why this isn't on the Q3 roadmap. You search "multi-region orders". NoteRank surfaces the original decision, the customer pattern, and a Q4 retro that already flagged the write-rate trend. The answer is two screens, not two meetings.

  4. Wed · 10:00

    Sprint retro

    The retro template runs through what shipped, what slipped, what's still open. AI Skills clusters the open items by the decision they trace back to. One pattern accounts for three of five. You log it.

  5. Thu · 14:15

    PRD draft

    You start the multi-region PRD. AI Co-Edit drafts the Background section with the customer voice from Tuesday, the retro insight from Wednesday, and the original Postgres decision from October. You edit, accept, edit. The first draft is forty minutes, not four hours.

  6. Fri · 16:00

    Stakeholder review

    You share a public link to the PRD. Comments land on the page, not in seventeen Slack threads. The Decision Log entry updates with a "review pending" status. By Monday, the call to ship has a record.

  7. Sun · 21:40

    Weekly review

    You write a one-paragraph review of the week — what shipped, what's still open, what you learned. Reflect & Crystals files it with thirteen other weekly reviews. In a quarter, you'll search "orders service" and see the whole arc.

None of this is theoretical. It's a sprint. The work is yours; the software keeps the connective tissue so you don't have to.

The Blind Spot — what every PM tool gets wrong

Product management is a memory problem, not a document problem.

The PM tool category solves one piece of the work and assumes the rest is your job. ChatPRD writes the PRD — and writes it well; it's the strongest tool in that lane. But the PRD is the answer; it doesn't tell you why you chose Postgres in October when engineering asks again in April. The decision lives outside the document.

Roadmap tools — Productboard, Aha!, Linear product views — manage the queue. They prioritize features and feedback. They don't remember the sprint where the strategy turned, the customer call that changed the prioritization, or the retro that flagged the write-rate trend two quarters before it became a problem. They track what you're shipping, not what you decided.

Generic note tools — Notion, Coda, Apple Notes — handle the capture. They're flexible enough that decisions can live there. The problem is what happens next: the note gets buried, search returns the wrong revision, AI on top sees rows and properties and not the connective tissue between a decision, a spec, a retro, and a customer voice.

Knovya is built for the connective tissue. Decision Log catches the choice. Experience Envelope groups it with everything shaped like it. NoteRank ranks the archive when the question returns. AI Co-Edit drafts with that whole context. The PRD still gets written. The roadmap still gets prioritized. Decisions stop evaporating between them.

The decision is the artifact, not the document around it. Build the artifact, and the document writes faster.

The Plan — for PMs, specifically

Three ways in. Pick the one your team needs.

Solo PMs run Pro. PM teams run Team. Free is enough to set up a decision log and see if the workflow holds for one sprint.

Free

$0 forever

Run one sprint through Knovya. Set up a decision log, capture a few retros, see if the recall feels different.

  • Up to fifty notes — one sprint of decisions and retros
  • Decision Log + Experience Envelope (read access to your own data)
  • AI Co-Edit, Conversation→Note — limited monthly credits
  • One public link, to share a single PRD with stakeholders
  • Templates for decision log, sprint retro, PRD, OKR
Open the workspace
For solo PMs

Pro

$15 per month

Built for the PM running their own decision archive. Unlimited notes, the full memory layer, the AI that drafts with context.

  • Unlimited notes — every sprint, every retro, every customer call
  • Full Decision Log + Experience Envelope + NoteRank precedent surfacing
  • Full AI Co-Edit, Conversation→Note, AI Transforms — credits scaled to PM work
  • Unlimited public links — share PRDs and decision pages with stakeholders
  • End-to-end encryption — competitive intel and unshipped strategy stay private
  • MCP for Claude, ChatGPT, Cursor — your decision archive is readable from any AI workflow
  • API and webhooks for Linear, Jira, Notion sync
Start with Pro
For PM teams

Team

$25 per seat / month

For PM groups running shared decision logs across squads. Real-time co-editing, shared folders, workspace permissions.

  • Everything in Pro, for the whole product team
  • Real-time co-editing on PRDs and decision pages
  • Shared folder structure with role-based permissions
  • Workspace-level Experience Envelope — org-wide decision precedent
  • Team templates and decision review workflows
  • Audit log and SAML SSO (enterprise add-on)
Start with Team

All plans include unlimited decision archive — your past decisions never expire on a free tier. See the full pricing →

Other Letters

We wrote six more. One might read like yours.

Try Knovya for the next sprint.

Bring last sprint's decisions, set up a decision log, run one retro. Free is enough to see if the recall holds.

Open the workspace Or browse all 26 features
Questions, answered

What PMs usually ask first.

  1. What are the best AI tools for product managers?

    The best AI tools for PMs split into three categories. Documentation tools (ChatPRD, Notion AI) write PRDs and specs. Roadmap tools (Productboard, Aha!) prioritize features. Analytics tools (Mixpanel AI, Amplitude) explain metrics. None of them remember why you decided what you decided. Knovya is built for that gap — a decision memory layer with sprint retros, customer interviews, and roadmap precedent in one place, surfaced through Experience Envelope and NoteRank.

  2. Is Knovya a ChatPRD alternative?

    Knovya solves a different problem than ChatPRD. ChatPRD writes PRDs — it's the strongest tool in that category. Knovya remembers why you chose Postgres in October when engineering asks again in April. Decision logs, sprint retros, customer interview synthesis, roadmap precedent — Knovya keeps the connective tissue. Many PMs run both: ChatPRD for the spec, Knovya for the context.

  3. How does Knovya help with decision logs?

    Decision Log templates capture every choice with the option you picked, the alternatives, the trade-offs, and the revisit threshold. Each decision auto-links to the spec, the meeting note, the ticket, and the retro. Six months later, search for the question and the original decision surfaces — with everything that informed it.

  4. Can Knovya synthesize customer interviews?

    Yes. Voice transcription turns a recorded interview into a structured note. AI Co-Edit summarizes themes across multiple interviews. Conversation→Note converts a Slack thread or a Claude session into a saved decision card. Customer feedback stops being thirty unread Notion pages and starts being a searchable, AI-rankable archive.

  5. Does Knovya integrate with Linear, Jira, or Notion?

    Knovya connects through MCP — the open standard that lets Claude, ChatGPT, Cursor, and other AI tools read your knowledge base directly. Linear, Jira, and Notion native integrations are on the roadmap; today, MCP-aware AI workflows can pull Knovya context into any client. API and webhooks are available for custom integrations.

  6. How is Experience Envelope useful for PMs?

    Experience Envelope groups past decisions by outcome — what worked, what failed, what's still open. When a question shaped like an old one arrives in a new sprint, the precedent surfaces itself. You don't dig for a decision you barely remember making; you see the pattern, the trade-off, and what shipped. The retro happens before the meeting.

  7. What's the difference between Knovya and Productboard?

    Productboard is a roadmap and feedback prioritization tool — it's the workhorse for PMs running a backlog of features and a stack of customer requests. Knovya doesn't manage the roadmap; it remembers the reasoning behind it. Past decisions, retros, customer voice, the call where the strategy turned. Many PMs use both: Productboard for the queue, Knovya for the why.