Methodology

1. Data collection

For each report theme, we identify 30 or so target companies and topic streams across the relevant geographies. Public sources include corporate IR pages, sustainability reports, regulator publications (EU Official Journal, FSA, SEBI, MoEFCC), industry press, and configured search feeds.

2. Triage

Collected documents are scored for relevance and classified by signal type (regulation, new product, M&A, supply capacity, financial, etc.) using theme-specific rules. Documents below a relevance threshold are dropped.

3. Synthesis

Per-vendor summaries are produced from structured source records and cited fact cards. Each synthesis captures an executive summary plus a key-events timeline that feeds the report's analytical sections.

4. Writing

Section drafts are built from the synthesized vendor profiles, theme-level context, and cited fact cards. Numeric claims are constrained to fact-card material — nothing is asserted beyond the cited sources.

5. Fact cards

Numeric claims and source citations are stored as structured "fact cards" with source URL, publisher, and published date. Public articles render selected fact cards inline; the full data layer is available with the report.

6. Delivery formats

Reports are delivered in three formats:

• PDF — fixed layout for reading and internal distribution
• Markdown — editable text, designed for reuse with LLM workflows
• JSONL — fact records in line-delimited JSON, suitable for direct ingestion into RAG pipelines or analytical tools

7. Limits and trust

We treat the reports as one input for decisions, not a complete substitute for primary due diligence. Sources are public and cited throughout. Material claims are anchored to public references rather than proprietary interviews.