dvcrn Company Research
Generate comprehensive Account Research Reports as professionally styled PDFs from a company URL.
Workflow
- Research the company (web fetch + searches)
- Build JSON data structure
- Generate PDF via
scripts/generate_report.py - Deliver PDF to user
Phase 1: Research (Parallel)
Execute these searches concurrently to minimize context usage:
WebFetch: [company URL]
WebSearch: "[company name] funding news 2024"
WebSearch: "[company name] competitors market"
WebSearch: "[company name] CEO founder leadership"
Extract from website: company name, industry, HQ, founded, leadership, products/services, pricing model, target customers, case studies, testimonials, recent news.
Phase 2: Build Data Structure
Create JSON matching this schema (see references/data-schema.md for full spec):
{
"company_name": "...",
"source_url": "...",
"report_date": "January 20, 2026",
"executive_summary": "3-5 sentences...",
"profile": { "name": "...", "industry": "...", ... },
"products": { "offerings": [...], "differentiators": [...] },
"target_market": { "segments": "...", "verticals": [...] },
"use_cases": [{ "title": "...", "description": "..." }],
"competitors": [{ "name": "...", "strengths": "...", "differentiation": "..." }],
"industry": { "trends": [...], "opportunities": [...], "challenges": [...] },
"developments": [{ "date": "...", "title": "...", "description": "..." }],
"lead_gen": { "keywords": {...}, "outreach_angles": [...] },
"info_gaps": ["..."]
}
Phase 3: Generate PDF
# Install if needed
pip install reportlab
# Save JSON to temp file
cat > /tmp/research_data.json << 'EOF'
{...your JSON data...}
EOF
# Generate PDF
python3 scripts/generate_report.py /tmp/research_data.json /path/to/output/report.pdf
Phase 4: Deliver
Save PDF to workspace folder and provide download link:
[Download Company Research Report](computer:///sessions/.../report.pdf)
Quality Standards
- Accuracy: Base claims on observable evidence; cite sources
- Specificity: Include product names, metrics, customer examples
- Completeness: Note gaps as "Not publicly available"
- No fabrication: Never invent information
Resources
scripts/generate_report.py- PDF generator (uses reportlab)references/data-schema.md- Full JSON schema with examples