Build Personas
The scenario
Your team keeps making assumptions about users. The existing personas are either too generic ("Marketing Mary") or buried in a Confluence page nobody reads. You need behavioral personas grounded in real research — personas that describe what users do, not just who they are — and you need them in a format the team will actually reference.
Example output
Persona: The Scaling Product Manager
Role: Senior Product Manager at Series B-D SaaS company (50-500 employees)
Jobs-to-Be-Done
- Standardize product management processes across a growing team
- Demonstrate product management impact to leadership with data
- Ship faster without sacrificing quality
Goals
- Reduce time spent on repetitive document creation (status updates, PRDs)
- Build a PM playbook the entire team can follow
- Get promoted to Director/VP by showing operational improvement
Frustrations
- "I spend more time formatting documents than actually thinking about the product"
- "Every PM on my team does things differently — there's no consistency"
- "I can't prove the value of PM to our CEO without better metrics"
Current Behavior
- Uses 4+ tools for product management work (Jira, Confluence, Notion, Slack)
- Writes PRDs from scratch every time — no templates or standards
- Sends status updates late or skips them when busy
Product Implications
- Needs templates that enforce consistency across the team
- Values time-saved metrics they can report upward
- Wants team-level features, not just individual productivity tools
Anti-Persona: The Freelance Product Manager
Solo consultant who values flexibility over consistency. Doesn't want structured workflows — prefers adapting their approach per client. Not our target user.
How to create this in 15 minutes
Step 1. Make sure your Product Management OS folder (second-pm-os) is open in Claude.
Step 2. In the chat panel, type:
/persona-generator
Step 3. Provide your research material. This works best when you give Claude real data to work with. You can:
- Point to interview snapshots in your
discovery/outputs/folder: "Use the 5 interview snapshots in discovery/outputs/" - Paste research notes, survey results, or behavioral data directly
- Describe what you know about your users if you don't have formal research yet
Step 4. Claude asks clarifying questions about your users and the segments you care about. Answer based on what you know.
Step 5. Claude generates behavioral personas and saves them to:
discovery/outputs/persona-[name].md
If something doesn't look right
Tell Claude what to change. For example:
- "The frustrations are too generic — make them more specific to our product"
- "Add an anti-persona for users who are NOT a good fit"
- "Split 'The Scaling Product Manager' into two personas — the team lead and the IC are very different"
- "Add a section about how this persona makes purchasing decisions"
- "These personas need to be based on the interviews, not assumptions — quote the research"
Pro tip
Run this after you've completed 3-5 interviews using the /user-interview-analyzer command. Feed the interview snapshots into /persona-generator and you'll get personas grounded in real evidence, not guesses. The more interviews you include, the more nuanced the personas.
What's next?
- Analyze All Your Research at Once — Process all your interviews into a unified synthesis
- Create a PRD — Write a PRD that references your personas automatically