Install the belt CLI skill:
npx skills add belt-sh/cli
Customer Persona
Create data-backed customer personas with research and visuals via inference.sh CLI.
Quick Start
Requires inference.sh CLI (
belt). Install instructions
belt login
# Research your target market
belt app run tavily/search-assistant --input '{
"query": "SaaS product manager demographics pain points 2024 survey"
}'
# Generate a persona avatar
belt app run falai/flux-dev-lora --input '{
"prompt": "professional headshot photograph of a 35-year-old woman, product manager, friendly confident expression, modern office background, natural lighting, business casual attire, realistic portrait",
"width": 1024,
"height": 1024
}'
Persona Template
┌──────────────────────────────────────────────────────┐
│ [Avatar Photo] │
│ │
│ SARAH CHEN, 34 │
│ Product Manager at a Series B SaaS startup │
│ │
│ "I spend more time making reports than making │
│ decisions." │
│ │
├──────────────────────────────────────────────────────┤
│ DEMOGRAPHICS │ PSYCHOGRAPHICS │
│ Age: 30-38 │ Values: efficiency, data │
│ Income: $120-160K │ Personality: analytical, │
│ Education: BS/MBA │ organized, collaborative │
│ Location: Urban US │ Interests: productivity, │
│ Role: Product/PM │ leadership, AI tools │
├──────────────────────────────────────────────────────┤
│ GOALS │ PAIN POINTS │
│ • Ship features │ • Too many meetings │
│ faster │ • Manual reporting (15 │
│ • Data-driven │ hrs/week) │
│ decisions │ • Stakeholder alignment │
│ • Team alignment │ is slow │
│ • Career growth to │ • Tool sprawl (8+ apps) │
│ Director │ • No single source of │
│ │ truth │
├──────────────────────────────────────────────────────┤
│ CHANNELS │ BUYING TRIGGERS │
│ • LinkedIn (daily) │ • Peer recommendation │
│ • Product Hunt │ • Free trial experience │
│ • Podcasts (commute) │ • Integration with Jira │
│ • Lenny's Newsletter │ • Team plan pricing │
│ • Twitter/X │ • ROI calculator │
└──────────────────────────────────────────────────────┘
Building a Persona Step-by-Step
Step 1: Research
Start with data, not assumptions.
# Market demographics
belt app run tavily/search-assistant --input '{
"query": "product manager salary demographics 2024 survey report"
}'
# Pain points and challenges
belt app run exa/search --input '{
"query": "biggest challenges facing product managers SaaS companies"
}'
# Tool usage patterns
belt app run tavily/search-assistant --input '{
"query": "most popular tools product managers use 2024 survey"
}'
# Content consumption habits
belt app run exa/answer --input '{
"question": "Where do product managers get their industry news and professional development?"
}'
Step 2: Demographics
Use ranges, not exact values. Personas represent a segment, not one person.
| Field | Format | Example |
|---|---|---|
| Age range | X-Y | 30-38 |
| Income range | $X-$Y | $120,000-$160,000 |
| Education | Common degrees | BS Computer Science, MBA |
| Location | Region/type | Urban US, major tech hubs |
| Job title | Role level | Senior PM, Product Lead |
| Company size | Range | 50-500 employees |
| Industry | Sector | B2B SaaS |
Step 3: Psychographics
What they think, value, and believe.
| Category | Questions to Answer |
|---|---|
| Values | What matters most to them professionally? |
| Attitudes | How do they feel about their industry's direction? |
| Motivations | What drives them at work? |
| Personality | Analytical vs intuitive? Leader vs collaborator? |
| Interests | What do they read/watch/listen to professionally? |
| Lifestyle | Work-life balance preference? Remote/hybrid/office? |
Step 4: Goals
What they're trying to achieve (both professional and personal).
Professional:
- Ship features faster with fewer meetings
- Make data-driven decisions (not gut feelings)
- Get promoted to Director of Product within 2 years
- Build a more autonomous product team
Personal:
- Leave work by 6pm more often
- Be seen as a strategic leader, not a ticket manager
- Stay current with industry trends without information overload
Step 5: Pain Points
Quantify whenever possible. Vague pain = vague persona.
❌ "Has trouble with reporting"
✅ "Spends 15 hours per week creating manual reports for 4 different stakeholders"
❌ "Too many tools"
✅ "Uses 8 different tools daily (Jira, Slack, Notion, Figma, Analytics, Sheets, Docs, Email) with no unified view"
❌ "Meetings are a problem"
✅ "Averages 6 hours of meetings per day, leaving only 2 hours for deep work"
Step 6: Jobs-to-be-Done (JTBD)
Three types of jobs:
| Job Type | Description | Example |
|---|---|---|
| Functional | The task they need to accomplish | "Prioritize the product backlog based on customer impact data" |
| Emotional | How they want to feel | "Feel confident presenting to the exec team" |
| Social | How they want to be perceived | "Be seen as the person who makes data-driven decisions" |
Step 7: Buying Process
| Stage | Behavior |
|---|---|
| Awareness | Reads blog posts, sees peer recommendations on LinkedIn |
| Consideration | Compares 3-4 tools, reads G2/Capterra reviews, asks in Slack communities |
| Decision | Requests demo, needs IT/security approval, evaluates team pricing |
| Influencers | Engineering lead, VP of Product, CFO (for budget) |
| Objections | "Will my team actually adopt it?", "Does it integrate with Jira?" |
| Trigger event | New quarter with aggressive goals, new VP demanding better reporting |
Step 8: Generate Avatar
# Match demographics: age, gender, ethnicity, professional context
belt app run falai/flux-dev-lora --input '{
"prompt": "professional headshot photograph of a 34-year-old Asian American woman, product manager, warm confident smile, modern tech office background, natural lighting, wearing smart casual blouse, realistic portrait photography, sharp focus",
"width": 1024,
"height": 1024
}'
Avatar tips:
- Match the age range, ethnicity representation, and professional context
- Use "professional headshot photograph" for realistic results
- Friendly, approachable expression (not stock-photo-stiff)
- Background suggests their work environment
- Business casual or industry-appropriate attire
The Anti-Persona
Equally important: who is NOT your customer.
ANTI-PERSONA: "Enterprise Earl"
- CTO at a 5,000+ person enterprise
- Needs SOC 2, HIPAA, on-premise deployment
- 18-month procurement cycles
- Wants white-glove onboarding and dedicated CSM
- WHY NOT: Our product is self-serve SaaS for SMB/mid-market.
Enterprise needs would require 2+ years of product investment.
Anti-personas prevent wasted effort on customers you can't serve.
Multiple Personas
Most products have 2-4 personas. More than 4 = too many to serve well.
| Priority | Persona | Role |
|---|---|---|
| Primary | The main user and buyer | Who you optimize for |
| Secondary | Influences the buying decision | Who you need to convince |
| Tertiary | Uses the product occasionally | Who you support, not target |
Validation
Personas based on assumptions are fiction. Validate with:
| Method | What You Learn |
|---|---|
| Customer interviews (5-10) | Real language, real pain points |
| Support ticket analysis | Actual problems, not assumed ones |
| Analytics data | Actual behavior, not reported behavior |
| Survey (50+ responses) | Quantified patterns across segments |
| Sales call recordings | Objections, buying triggers, language |
Common Mistakes
| Mistake | Problem | Fix |
|---|---|---|
| Based on assumptions | Fiction, not research | Start with data |
| Too many personas (6+) | Can't serve everyone well | Max 3-4 |
| Vague pain points | Not actionable | Quantify everything |
| Demographics only | Misses motivations and behavior | Add psychographics, JTBD |
| Never updated | Becomes outdated | Review quarterly |
| No anti-persona | Wasted effort on wrong customers | Define who you're NOT for |
| Single persona for all | Different users have different needs | Primary/secondary/tertiary |
Related Skills
npx skills add inference-sh/skills@web-search
npx skills add inference-sh/skills@ai-image-generation
npx skills add inference-sh/skills@prompt-engineering
Browse all apps: belt app list