The problem this solves
Who a company works with explains what it sells and who it sells to better than any description it writes about itself. The customer logos on its homepage, the partners on its integrations page and the case studies in its resources section are a public map of its market, and that map is useful three ways: to size the company, to find lookalike prospects among its customers, and to find partnership routes through the companies it already integrates with.
Reading that map by hand is slow because it is spread across pages, hidden in logo walls with no text, and full of brand names that need resolving. The Company Intelligence Agent crawls the site, reads the pages as a visitor sees them, resolves every named company to a domain with firmographics, and lays the relationships out in a sheet with the page each one came from.
How the mission runs
- Crawl the site for relationship pages. The site crawl maps <company.com> and identifies the pages that name other companies - customers, case studies, partners, integrations, marketplace listings - and the rendered browser reads pages where logos load by script so nothing on a logo wall is missed.
- Extract and classify the names. Every named company is extracted with its relationship type - customer, partner, integration - and the page it appeared on. Public sources such as press releases are checked for partnerships not shown on the site, with the source recorded.
- Resolve and enrich each company. Enrichment resolves each name to a domain and adds headcount, industry and HQ, with ambiguous names resolved using the context of the page and flagged where uncertainty remains.
- Write the relationship map. Google Sheets receives one row per relationship with the company, its domain, relationship type, firmographics and the source page link, grouped by type, with a summary tab counting relationships by type and industry.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A Google Sheet mapping the customers, partners and integrations of <company.com>, one row per relationship with the resolved domain, headcount, industry, HQ, relationship type and the source page or press release it came from, plus a summary tab of counts by type and industry. Unresolved names are listed with the agent's best candidate marked.
Make it yours
- Turn the customer list into a lookalike search to find prospects that resemble the company's own customers.
- Run it for three competitors and compare their customer and partner overlap.
- Add the partnerships contact at each partner with a verified email for a channel program.
- Save the map to Notion instead of Sheets.
- Refresh quarterly and highlight relationships added or removed since the last map.
Frequently asked questions
How does it read logo walls with no text?
The rendered browser captures the page as displayed and the agent identifies logos from image names, links and alt text, resolving each to a company and flagging any it could not identify.
Are the relationships current?
They reflect the site and public sources at run time. Case studies can be old; the source page and any visible date are recorded so you can judge freshness.
Can it estimate how important a customer is to them?
It notes prominence - homepage placement, a dedicated case study - as a proxy, without claiming revenue figures the sources do not provide.
Can it find relationships the site does not show?
Press releases, partner directories and marketplace listings are checked as public sources, and each relationship found that way carries its source so it can be distinguished from the site's own claims.
Does it read the integrations marketplace too?
Yes. Where the company runs a marketplace or integrations directory, each listed integration is captured with its category, which is often the most complete public statement of the company's ecosystem.