Claude Research Synthesis Pack — Goodbye Gartner

I want to run market research in Claude instead of waiting six weeks on a Gartner seat. Walk me through it.

This is NOT a Terminal or coding install. The prompts run inside the Claude app. Treat me like a strategy lead, product lead, or founder who has never coded. The only optional technical bit is a free public-data source, with a no-install fallback.

What I am setting up: a private Claude Project that holds my question and any sources, so the vault on this page can scan the market, build the vendor wave, size the TAM, and draft the board brief, on my exact question, this week. A named human signs the brief before it informs a decision.

Walk me through ONE step at a time. Wait for me to confirm each.

1. Open `claude.ai`, sign in, pin Opus 4.8. For confidential strategy work use Team or Enterprise so it stays in my tenant.

2. Left sidebar → "Projects" → "Create Project". Name it for the question (for example "Vendor X Evaluation" or "Category Sizing").

3. Load what I have into Project Knowledge: prior reports, vendor decks, filings, my own notes. This is data source (A) in prompt 01. If I have nothing yet, that is fine, the desk pulls public signals (option B).

4. Optional, live signals and financials. For public financials, OpenBB (`github.com/OpenBB-finance/OpenBB`) is a free open-source data layer; for broad web/news gathering, an open-source deep-research agent like GPT-researcher (`github.com/assafelovic/gpt-researcher`) can pull signals; or I just paste links and let Claude read them. Skip if my Project sources are enough.

5. Set Custom Instructions: "You are a research desk supervised by a named human. Every claim cites a real source or is labeled an estimate with method. Never invent a market size, vendor capability, or quote. A human signs the brief before it informs a decision."

6. Run prompt 01 (the router): pick what I need (scan, wave, sizing, teardown) and frame the exact question. Then run 02 to 10: scan, vendor wave, sizing, signals, procurement score, teardown, tracker, board brief, source-tie audit.

7. Confirm the output bar: every claim sourced or labeled an estimate, prompt 10 blocks any unsupported claim before the brief ships.

## Rules for walking me through this

- One step at a time. Tell me exactly what to click or paste.
- Define jargon once: Project, Project Knowledge, Custom Instructions, vendor wave, TAM/SAM/SOM, `<review_gate>`, `{{TOKEN}}`.
- Never tell me to use a Terminal for the prompts. The vault runs from a browser.
- Anti-pattern call-outs: if my firm blocks claude.ai, use Claude Enterprise with SSO; if I have no sources, lean on the public-signal prompts (05) and label estimates clearly; do NOT tell me a step is "not possible", give me the paste fallback.
- A named human signs the brief. Claude researches and ties sources; a person decides.

First message: ask me "What do you need, a market scan, a vendor wave, a market size, or a competitor teardown? And what is the exact question and the decision deadline?" Then start step 1.

When prompts 01 to 03 run on my real question and the wave holds up, switch into "desk mode" and help me set up the recurring tracker so the category stays live.