Skip to main content
These examples show what you can ask once Evermuse is connected, and how a well-behaved agent uses the tools to answer. The exact tool sequence varies by client and model. Every answer cites its evidence with links that open the original meeting, quote or document in Evermuse.
You don’t need to name tools or ids. Mention the product if your workspace has more than one, and the agent will find the rest.

​
Research customer pain points

Prompt

“What are the biggest pain points in our customer interviews from the last quarter?”
What happens
  1. find_skills and read_skills load the research methodology (first conversation only).
  2. get_products identifies the product. The agent asks you to pick one if there are several.
  3. customer_research runs the first evidence search and returns the customer-research method.
  4. search in browse mode (note_types: ["problem"] with date_from set to three months ago) pulls recent pain points. view_item opens the most important ones for full context.
What you get: a ranked list of pain-point themes. Each has how many distinct customers raised it, representative verbatim quotes, and inline citations linking to each quote in Evermuse. Gaps where the evidence is thin are called out.

​
Draft a PRD from customer evidence

Prompt

“Use our research to draft a PRD for improving onboarding.”
What happens
  1. get_products and get_product_summary give the agent the product’s goals, users and existing features.
  2. create_prd searches for onboarding evidence and returns the PRD methodology and template.
  3. Follow-up search calls fill gaps, for example evidence on activation and setup time. find_sources and read_source pull the full context of key interviews.
What you get: a business PRD: problem, target users, evidence-backed requirements, success metrics and open questions. Every requirement cites the customer evidence behind it. Ask “Save this as a shaping note” and the agent calls create_shaping_note to store it in Evermuse.

​
Summarize a customer call

Prompt

“Summarize my call with Acme from last Tuesday and list the follow-ups we promised.”
What happens
  1. summarize_conversation searches sources with participant: "@acme.com" and a date range, and returns the summarization method.
  2. read_source reads the transcript, paging through long calls.
What you get: a structured summary with attendees, key discussion points, customer needs and objections, and the commitments made, each linked to the moment in the transcript.

​
Compare against competitors

Prompt

“How do we compare to our top competitors, and where are customers telling us we fall short?”
What happens
  1. competitor_analysis searches customer mentions of competitors and returns the competitor-analysis method.
  2. list_competitors lists tracked competitors, and get_competitor_capabilities fetches recent releases and capabilities for the most relevant ones.
  3. search finds customer evidence about switching, comparisons and missing features.
What you get: a comparison of capabilities, the gaps customers actually mention (with quotes), and differentiation opportunities.

​
Prioritize what to build next

Prompt

“Which opportunities have the strongest customer evidence right now?”
What happens
  1. get_opportunities lists suggested opportunities with evidence counts and the number of distinct companies behind each, and returns the prioritization method.
  2. view_item with item_type: "opportunity" opens the strongest candidates, and search checks for counter-evidence.
What you get: a ranked shortlist with reach (distinct customers), the quotes behind each opportunity, and the risks or open questions to resolve.

​
Write user stories for a feature

Prompt

“Write user stories with acceptance criteria for bulk CSV export.”
What happens: user_stories searches evidence about exporting data and returns the user-story method. The agent follows up with search for edge cases customers have hit. What you get: INVEST user stories, each with acceptance criteria tied to real customer requests and the edge cases customers reported.

​
Daily brief

Prompt

“Give me a brief of what we learned from customers since yesterday.”
What happens: write_brief searches recent evidence and returns the brief format. find_sources lists the meetings and threads processed in that window. What you get: a short brief of new themes, notable quotes, and changes compared with earlier evidence. It quotes customers only, never your internal team.

​
Save research back to Evermuse

Prompt

“Add these interview notes to the Onboarding project in Evermuse.” (with notes pasted or a file attached)
What happens
  1. get_projects finds the Onboarding project.
  2. add_source queues the notes (or uploaded document) for processing, with nature: "evidence".
What you get: confirmation that the source was queued. When processing finishes, Evermuse extracts needs, feedback and quotes from it, and they show up in future searches.
add_source, add_signals, update_signals, create_shaping_note and update_shaping_note change your workspace. Most clients ask you to confirm before running a tool that isn’t read-only.

​
Run a skill directly

In clients that support MCP prompts, you can run any Evermuse skill as a slash command. For example, in Claude Code:
See Skills & prompts for the full list.