If you look at companies often, doing it ad hoc means you forget steps and lose your notes. A light system, a saved prompt, a consistent template, and a source-checking habit, makes every next company faster and more reliable.
What you'll have by the end: a reusable research prompt, a consistent place to store findings, and a routine that keeps facts separate from guesses.
1. One saved prompt, every time
Use the same prompt for every company so you get comparable answers. This one asks the AI to tag how solid each claim is, which is the whole game:
Research [COMPANY] like an investor. For EVERY claim, tag it:
[FACT] (with source + date), [COMPANY CLAIM] (from them, unverified),
or [ESTIMATE] (reported, not confirmed). If unknown, say "unknown".
Cover: what it sells, who pays and the pricing, how it reaches
customers, what makes it defensible, and what could stop it,
including which AI model it depends on, its rough margins, and any
legal or regulatory risk. End with what you couldn't find.2. A consistent place to store it
Keep one document or table per company, always in the same shape: what it does, who pays, the numbers (with sources and dates), your read, and open questions. Sameness is what lets you compare later.
3. Always check the AI's sources
Treat the AI's first draft as a starting point, not the answer. Open the sources it cites, verify the few numbers that matter, and downgrade anything you can't confirm to "estimate". This is the step that separates research from vibes.
4. Keep it fresh
Note the date you researched each company. This space moves fast, funding, pricing, and legal status change, so a six-month-old note is a lead, not a fact.
For the full framework and a worked example, see research a business before you invest.