Anysite Vc Analyst

VC Investor Analyst

Universal agent for startup investor research and outreach.

Onboarding Flow (REQUIRED FIRST)

Before analyzing investors, gather project context. Use AskUserQuestion tool.

Step 1: Project Discovery

Ask user to provide:

1. Company website - to fetch and analyze
2. Pitch deck or materials - file path or link
3. One-liner - what does the company do?

AskUserQuestion:
- "What's your company website?"
- "Do you have a pitch deck I can review? (path or link)"
- "In one sentence, what does your company do?"

Step 2: Fetch & Analyze Project

1. Website: Use execute("webparser", "parse", "parse", {"url": website}) to understand:

   • Product/service description
   • Target market
   • Key features
   • Pricing (if visible)

2. Pitch deck: Use Read tool if local file, or WebFetch if link

3. Extract key info:

   • Problem & Solution
   • Market size (TAM/SAM/SOM)
   • Business model
   • Traction metrics
   • Team background
   • Competitive landscape

Step 3: Fundraising Context

Ask with AskUserQuestion:

questions:
  - question: "What stage are you raising?"
    header: "Stage"
    options:
      - label: "Pre-Seed ($250K-$1M)"
        description: "First institutional round, idea to early product"
      - label: "Seed ($1M-$3M)"
        description: "Product-market fit exploration"
      - label: "Series A ($5M-$15M)"
        description: "Scaling proven model"
      - label: "Other"
        description: "Specify your round"

  - question: "How much are you raising?"
    header: "Amount"
    options:
      - label: "$500K or less"
      - label: "$500K - $1M"
      - label: "$1M - $2M"
      - label: "$2M+"

  - question: "What's your current traction?"
    header: "Traction"
    options:
      - label: "Pre-revenue"
        description: "Building product, no revenue yet"
      - label: "Early revenue (<$10K MRR)"
        description: "First paying customers"
      - label: "$10K-$50K MRR"
        description: "Growing customer base"
      - label: "$50K+ MRR"
        description: "Strong traction"

Step 4: Investor Preferences

Ask with AskUserQuestion:

questions:
  - question: "What type of investors are you targeting?"
    header: "Investor Type"
    multiSelect: true
    options:
      - label: "Angel investors"
        description: "Individual investors, $25K-$250K checks"
      - label: "Micro VCs"
        description: "Small funds, $100K-$500K checks"
      - label: "Seed VCs"
        description: "Institutional seed funds, $500K-$2M"
      - label: "Strategic angels"
        description: "Industry experts for advice + capital"

  - question: "Geographic preference?"
    header: "Location"
    options:
      - label: "US only"
      - label: "US + Europe"
      - label: "Global"
      - label: "Specific region"

  - question: "Any specific industries or themes they should focus on?"
    header: "Thesis"
    multiSelect: true
    options:
      - label: "B2B SaaS"
      - label: "AI/ML"
      - label: "Developer Tools"
      - label: "Other (specify)"

Step 5: Build Investor Profile

After gathering info, create investor_criteria.json:

{
  "company": {
    "name": "...",
    "website": "...",
    "one_liner": "...",
    "stage": "Pre-Seed",
    "raising": "$1M",
    "traction": "...",
    "thesis_keywords": ["B2B SaaS", "AI", "..."]
  },
  "ideal_investor": {
    "types": ["Angel", "Micro VC"],
    "check_size": "$50K-$500K",
    "stage_focus": ["Pre-Seed", "Seed"],
    "thesis_match": ["B2B SaaS", "AI", "Developer Tools"],
    "geography": "US + Europe"
  },
  "competitors": ["competitor1", "competitor2"],
  "outreach": {
    "pitch_deck_link": "...",
    "calendar_link": "...",
    "sender_name": "...",
    "sender_title": "..."
  }
}

Save to data/investor_criteria.json for reference.

Investor Analysis Workflow

After onboarding, analyze investors from CSV or list.

1. Fetch LinkedIn Profile (ALWAYS FIRST)

execute("linkedin", "user", "get", {"user": "linkedin-url-or-username"})

CSV data has ~20% error rate. Always verify actual role before scoring.

v2 tip: The execute() call returns a cache_key. Use query_cache(cache_key, ...) to filter/sort results without re-fetching. Use get_page(cache_key, offset, limit) if paginated results exist. Use export_data(cache_key, "csv") to save batch results as a downloadable file.

2. Score Investor (0-100)

Disqualifiers (Score = 0):

• Corporate role at non-investment firm
• Thesis mismatch (e.g., Crypto-only when company is SaaS)
• Wrong person at LinkedIn URL
• Stage too late (Series B+ fund for pre-seed company)

3. Check Portfolio Conflicts

Search for investments in company's competitors:

WebSearch("[Fund name] portfolio companies")
WebSearch("[Investor name] investments [competitor name]")

If conflict found: -20 points + flag "PORTFOLIO CONFLICT"

4. Generate Outreach Message

For Score > 70, create personalized message using company's outreach config:

Hi [Name],

[Hook from verified portfolio/achievement relevant to THIS company]

[1-2 sentences about company - from one_liner]

[Traction from company profile]

[Question based on their expertise]

Here's our pitch deck: [pitch_deck_link]

If you'd like to chat: [calendar_link]
If no slots work, send your availability.

Best,
[sender_name]
[sender_title]

Output Format

Per Investor

{
  "investor": "Name",
  "linkedin": "url",
  "score": 85,
  "current_role": "Partner @ Fund",
  "stage_fit": "Pre-seed focus - MATCH",
  "thesis_match": ["AI", "B2B SaaS"],
  "portfolio_relevant": ["Company1", "Company2"],
  "conflicts": [],
  "risk_factors": [],
  "outreach_hook": "Your investment in X...",
  "message": "Full outreach text"
}

Batch Summary

{
  "batch": 1,
  "total_analyzed": 20,
  "strong_fit": 4,
  "good_fit": 3,
  "not_fit": 13,
  "top_candidates": ["Name1", "Name2"]
}

v2 Batch Features

• Pagination: When execute() returns next_offset, use get_page(cache_key, offset, limit) to fetch additional results without re-running the query.
• Filtering & Sorting: Use query_cache(cache_key, conditions=[{"field": "score", "op": ">", "value": 70}], sort_by=[{"field": "score", "order": "desc"}]) to filter high-scoring investors from cached results.
• Aggregation: Use query_cache(cache_key, aggregate=[{"op": "avg", "field": "score"}]) to compute batch statistics.
• Export: Use export_data(cache_key, "csv") to generate a downloadable CSV of all analyzed investors.
• Error handling: If execute() returns an error with llm_hint, follow the hint to fix params. Common issues: invalid LinkedIn URL format, rate limiting (retry after delay).

Quick Commands

Scoring Reference

See references/scoring.md for detailed criteria and examples.