Search both a company's own disclosures and other issuers' references to it. The result is an evidence-backed relationship map that can help investigate customer dependence and supply-chain exposure.

## The flow

1. **Set the company, aliases, and time window.** Find the company's legal name, relevant subsidiaries, brands, and products in its source filings. Keep each identity explicit rather than treating every brand reference as a statement about the entire parent. State the research end date and look-back window.
2. **Search in both directions.** Use ticker-scoped [SearchDocuments](/docs/mcp/tools/filings#searchdocuments) for the company's customer and supplier disclosures. Then omit `ticker` to search other issuers for its name and supported aliases, using separate short queries. Date filters bound the source filing or call dates. `maxResultsPerCompany` spreads discovery across issuers; `excludeTickers` can remove the starting company only in the market-wide pass. Search available filings and transcripts and record which sources were examined.
3. **Establish the relationship from the actual passage.** Use [SearchDocument](/docs/mcp/tools/filings#searchdocument) with `searchMode: "exact"` to locate a name, then [ReadDocumentLines](/docs/mcp/tools/filings#readdocumentlines) to read the context. Record who supplies whom, what is supplied, and whether the statement describes a current relationship, a past relationship, or a proposed agreement. Classify customers, suppliers, distributors, partners, and competitors separately. An incidental mention or analyst question remains unconfirmed. Corroborate with the counterparty's disclosures where available and preserve conflicting or differently dated statements.
4. **Add disclosed financial exposure.** [GetCustomerConcentration](/docs/mcp/tools/fundamentals#getcustomerconcentration) returns supported customer-concentration disclosures; pass `maxFilings` greater than one to inspect earlier filings. Preserve the percentage, revenue or receivables basis, exact period, customer count or identity as disclosed, and source quote. Leave unnamed customers unnamed. Do not attach an anonymous concentration percentage to a named relationship discovered elsewhere, and do not treat missing coverage as zero exposure. Supplier spending or dependence requires its own source disclosure.
5. **Connect the map to an economic question.** Use [GetRevenueBreakdown](/docs/mcp/tools/fundamentals#getrevenuebreakdown) for annual business context, and relevant [GetGuidance](/docs/mcp/tools/fundamentals#getguidance) or [GetCompanyKpis](/docs/mcp/tools/fundamentals#getcompanykpis) for disclosed demand and spending signals. A spending change at a customer can motivate a supplier scenario; quantify the impact only when the disclosed relationship, allocation, periods, and assumptions support it. Show estimates as scenarios and leave undisclosed allocations unknown.

## Prompt

> Map NVIDIA's disclosed customers, suppliers, distributors, and partners using its filings and references to NVIDIA and its relevant products in other companies' filings and available transcripts from the past 12 months. For every relationship, show the two counterparties, supply direction, product or service, relationship type, current/past/proposed status, who made the statement, a short quote, source date, and source link. Separate confirmed disclosures from unresolved mentions. Add customer-concentration percentages only where the source supports that specific identity, basis, and reporting period; retain anonymous customer disclosures separately. Explain which relationships could transmit a customer spending change, and show missing exposure figures and coverage gaps without estimating undisclosed allocations.

## What you get

A dated relationship table with source evidence and a separate view of disclosed financial dependence. It gives the model a foundation for researching supply-chain scenarios without inventing counterparties or exposure percentages.

Use [Discover a company's competitors](/docs/recipes/discover-competitors) to investigate rivals, or [Compare earnings calls over time](/docs/recipes/compare-earnings-calls) to track a counterparty's changing demand outlook.