Company Tearsheet
A disciplined one-pager from whatever materials you have.
Claude: upload the .zip under Settings → Capabilities → Skills. Claude Code: copy SKILL.md into a folder named “company-tearsheet”. ChatGPT Skills: upload the .zip from Skills → New skill where enabled by your plan and workspace; otherwise paste SKILL.md into a Custom GPT's instructions or a Project.
name: company-tearsheet description: Builds a one-page company tearsheet or profile from provided filings, decks, transcripts and notes, with strict source discipline. Use when the user asks for a tearsheet, one-pager, company profile or quick summary of a business.
Compress the user's materials on a company into a disciplined one-page profile. The value is consistency and sourcing: every tearsheet has the same shape so names are comparable side by side and every fact is traceable. A tearsheet gets pulled up thirty seconds before a meeting. It has to be scannable and it has to be right.
Inputs
- Any materials on the company: filings, investor decks, transcripts, sell-side notes, internal notes.
- Optional: the purpose. Screening, meeting prep, IC appendix, watchlist maintenance. This sets the depth of each section (screening favors the KPI table; meeting prep favors recent developments and open questions).
Output format: one page
Header: company name, one-line descriptor, HQ, listing/ownership status and the as-of date of the newest source material, stated plainly so staleness is visible: "Sources as of: Q3 FY25 10-Q (Nov 2025)."
Business model: how it makes money in three or four sentences: what is sold, to whom, the revenue model (recurring vs. Transactional, contract structure, pricing basis) and segment mix with percentages if disclosed.
Key figures: a compact table: revenue and growth, gross margin, EBITDA or operating margin, leverage and two or three sector-appropriate KPIs. Choose KPIs the way an analyst in that sector would: NRR and gross retention for software; same-store sales and unit growth for retail; combined ratio for insurers; occupancy and rate for real assets; AUM flows and fee rate for asset managers. Label the period and source on every figure.
| Metric | Value | Period | Source |
|---|---|---|---|
| Revenue | $412M (+18% y/y) | FY25 | 10-K p.44 |
| Net revenue retention | 114% | Q3 FY25 | Investor deck p.9 |
Ownership and capital structure: major holders, insider ownership, debt picture (amount, maturity wall, rate exposure) and anything unusual: dual-class shares, converts, large warrants, activist presence.
Management: CEO and CFO with tenure; note recent turnover in either seat, which is always worth a line.
Recent developments: the three to five most material items, dated, one line each, most recent first.
Bull / Bear: the two or three strongest points on each side, drawn from the materials, written as arguments rather than facts ("Bull: pricing reset in FY26 as legacy contracts roll. 60% of book renews within 18 months, deck p.12").
Open questions: what the materials do not answer, phrased as the questions an analyst would actually ask next.
Workflow
- Read everything provided, then fill the template in order.
- Every figure carries a source label. Document and page or section where possible.
- Where the materials are silent, write "[not in provided materials]". Do not fill gaps from general knowledge.
- If the user explicitly asks you to supplement from general knowledge, keep those additions in a separate, clearly labeled block and remind them to verify before use.
- If the provided materials would leave more than a third of the template blank, say so up front and offer the labeled general-knowledge supplement before delivering a hole-riddled page.
Guardrails
- Never blend remembered figures with sourced figures. A tearsheet with silent guesses is worse than one with visible holes. The reader cannot tell which numbers to trust, so they can trust none.
- Keep the format identical across uses; comparability is half the value.
- No investment recommendation. The tearsheet describes; the reader decides.