Muse for Product Managers

A product manager’s day is mostly triage. Slack threads that grew overnight. An inbox where a customer escalation sits two rows above a vendor newsletter. A calendar with seven meetings and no indication of which one needs preparation. The actual product work happens in the cracks between these, if there are any cracks left.

I’ve run my week with Muse as a chief of staff for months now. The real change is that Muse reduces the day to an attention list: the handful of things that need my judgment, surfaced with enough context that I can decide in minutes what used to take an hour of rummaging. That is the whole job of this lesson: fewer tabs.

The attention list: Slack, email, calendar, one page

The core setup has a name: the Stack-Rank. It is a scheduled routine, weekday mornings, that reads my Slack mentions and key channels, my inbox, and my calendar, then produces one page: every item ranked by what actually needs me today, each with a one-line why and a suggested action.

Getting this right is a setup problem. Muse needs the connectors: Slack, Gmail, and Google Calendar. Then the routine needs the ranking logic, because the default instinct of any agent is to rank by loudness. A heated Slack thread outranks a quiet calendar prep item every time, even when the meeting is the thing that matters. So the brief has to say so explicitly. Mine looks like this:

Every weekday at 6:30am, build my attention list. Read my Slack mentions and the #eng, #design, and #product channels; scan Gmail for anything from the exec team, customers, or vendors; read today’s calendar. Rank by: 1) things where someone is blocked waiting on me, 2) decisions due before noon, 3) meetings today that need prep, 4) everything else worth a glance. For each item: what it is, why it made the list, the action I should take, and how long it will take. Cap the list at ten items. Do not draft replies or reschedule anything, just the list.

The specialist pattern applies here. The Stack-Rank reads and ranks; it does not send. Replies, calendar moves, and Slack posts wait for my approval one at a time, because the lesson on approval lines applies double to an agent that reads everything I read.

Research synthesis: conclusions with receipts

Product managers drown in unstructured input: interview notes, survey verbatims, support tickets, competitive teardown threads, analyst reports. The failure mode is asking Muse to “summarize the feedback,” which produces a confident paragraph that flattens five distinct problems into one smooth theme. Themes are where the real signal goes to die.

The practitioner version works like this. Feed Muse the raw material, then ask for extraction: every distinct complaint or request, how many times it appeared, who said it, and a quote. Themes come second, and only from the extracted items. The prompt that runs this is below, and the non-negotiable part is the proof requirement: cited, checkable output. If a claim does not trace to a line in the source material, it does not go in the brief. My rule is blunt: any research output Muse gives me without sources gets deleted.

The pre-call brief: thirty seconds, before every meeting

This is the highest-leverage routine I run. Before each meeting, Muse produces a half-page: who is on the call and what they care about (from memory and prior threads), the open threads with each stakeholder, the last decision we made together, and the one question this meeting needs to answer. The calendar connector tells it which meeting is next; standing memory supplies the stakeholder context; recent Slack threads supply the open loops.

The morning triage routine and the pre-call brief can be one job or two. I keep them separate: the triage runs at 6:30am on a schedule, and the pre-call brief is on-demand, one message before I join. Separating them means each prompt is short, which keeps the output focused. Short prompts are load-bearing. A prompt that tries to do triage, briefs, and follow-up drafting in one go produces a slurry. One job, one output.

The copy-paste prompts

Prompt 1: The Stack-Rank morning brief

Paste this into Muse after connecting Slack, Gmail, and Google Calendar:

Build my morning attention list. Check my Slack mentions and the #[eng/product/design channel names] channels, scan Gmail for anything from executives, customers, or vendors, and read today’s calendar. Rank items by: 1) anyone blocked waiting on me, 2) decisions due before noon, 3) meetings today that need prep, 4) everything else worth a glance. For each item give me: what it is, why it made the list, the action I should take, and an honest time estimate. Cap at ten items. Read only, do not draft anything or change anything.

Prompt 2: Research extraction with receipts

Here is [paste interview notes / survey exports / ticket dumps]. Extract every distinct complaint, request, or observation. For each one: the exact quote, who said it, and how many times the same point appeared across the material. Then, and only then, group the extracted items into themes, each theme citing the items it came from. No conclusions, no recommendations, no invented quotes. If a quote cannot be traced to the text, leave it out.

Prompt 3: The pre-call brief

I have a meeting with [name/role] at [time]. Give me a half-page brief: what they care about based on our past conversations, any open threads I have with them from Slack or email, the last decision we made together, and the single most important question this meeting should answer. Facts only, from my connected apps and your memory. If you are unsure about something, say so.

What can go wrong: the urgency trap

Here is the honest failure mode. A well-tuned attention list optimizes for what demands attention, and the loudest things are rarely the most important. Week by week, the Stack-Rank will drift toward fires: escalations, blocked engineers, heated threads. The quiet work, roadmap thinking, writing the one-pager, the uncomfortable stakeholder conversation you keep postponing, will quietly slide off the list, because it never generates a Slack mention or an urgent email. An attention list that only ranks loudness is a very efficient way to stop doing product strategy.

The fix is embarrassingly manual. Once a week, ask Muse to audit the last seven attention lists and name what never appeared. Then give it standing instructions that survive in memory: “always include one strategic item, even if nothing in Slack or email supports it.” And keep one thing the routine can never schedule for you: the judgment itself. Muse assembles the agenda. You still have to notice what is missing.