Investor research

How it works#

1. Identity first. A cheap search works out who the investor is professionally (LinkedIn, firm website, title). A clear match is saved; an unclear one shows “Is this them?” with two or three candidates, and nothing deeper runs until you pick. A LinkedIn URL you entered yourself is trusted without a search.
2. Search. Three searches read recent deals, posts and talks, and the fund. Pages that do not match the confirmed identity, other people's profiles and personal social sites are thrown away.
3. Write. The writer model reads the pages that were kept and writes the dossier. Every fact cites one of those pages; what it cannot cite goes to a collapsed “Unverified” section.
4. Check. Raisebook checks every citation again and strips phones, addresses and private topics before anything is saved.

Empty investor fields (check size, stage focus, title…) are filled and marked “from research”. A field you already filled is never changed: research suggests the new value, and you accept or ignore it, one at a time or all at once (A in Triage).

Presets and cost#

Settings → AI research shows the estimate per dossier for each preset from OpenRouter's current prices, usually cents. Matching an identity costs about a cent. Advanced lets you pick any OpenRouter model for each step. The cost OpenRouter reports for each call is recorded on the run.

The monthly limit#

Each workspace has a monthly limit, $25 by default; owners change it. Before every model call Raisebook sets aside that call's worst case under the limit, and settles the real cost afterwards, so parallel runs never go over it. A run that reaches the limit stops cleanly, keeps what it gathered, and can be resumed. Bulk research shows the count, who is left out and the total before you confirm, and offers the part that fits.

Keeping dossiers fresh#

• A dossier older than 90 days shows “May be stale”.
• Moving someone to Meeting with a dossier older than 30 days offers “Refresh before your meeting?”. It never runs by itself.
• A refresh shows “New since” with what changed at the top, and the last five versions are kept.

Agents: save_dossier#

An agent such as Claude Code can research with its own web tools and save the result with save_dossier (or POST /v1/investors/{id}/dossier). It needs no OpenRouter key and costs the workspace nothing. The dossier goes through the same checks as a model's and is shown “by claude-code”. Facts without a source the agent fetched are dropped and listed in the answer.

{  "schema_version": 1,  "identity": {    "name": "Dana Whitfield",    "firm": "Harbor Light Ventures",    "title": "Partner",    "linkedin_url": "https://www.linkedin.com/in/example-dana-whitfield",    "firm_website": "https://harborlight.example"  },  "sources": [    { "id": "s1", "url": "https://harborlight.example/news/brightquay-seed",      "title": "Harbor Light leads Brightquay's seed", "published_at": "2026-08-14" }  ],  "facts": [    { "key": "f1", "section": "activity", "kind": "deal", "date": "2026-08-14",      "text": "Led the $3M seed round in Brightquay Robotics.", "source_ids": ["s1"] }  ],  "analysis": {    "brief": {      "verdict": "strong",      "why": "Leads seed rounds in robotics at our check size.",      "best_way_in": "Brightquay's founders, whose round she led.",      "opener": "Saw your Brightquay seed; we automate the same warehouses.",      "risk": "May see us as close to Brightquay.",      "fact_keys": ["f1"]    },    "ways_in": [{ "text": "Brightquay's founders", "basis": "public", "fact_keys": ["f1"] }],    "talking_points": [{ "text": "Lead with the warehouse pilots.", "fact_keys": ["f1"] }]  },  "suggestions": { "check_min": { "value": 250000, "fact_keys": ["f1"] } }}

Sections: profile, fit, activity, thesis, red_flags. The other MCP tools are get_dossier, research_investor, match_identity and list_research_runs. MCP tools reference.

Privacy#

Research sends the investor's name, firm, title and links, your company profile and the round's terms to OpenRouter. It never sends notes, emails, amounts or timelines. Dossiers hold professional information only and stay private to your workspace. Investors can ask for removal through the Research Policy.