Company lookalike API
lookalikes
2.5 credits / company returnedFind the companies most similar to one or more seed companies by meaning (embedding similarity over the company graph). Pass a single seed OR an array of seeds (domain, company name, or org_id) — multiple seeds blend into one centroid so you get neighbors similar to the SET, ideal for expanding a target list from your best accounts. Optionally narrow by industries, company_size, and locations. Enable `rerank` for Smart mode (LLM business-model rerank: higher precision, slower). Paginate with limit/offset (the similarity neighborhood is bounded: offset+limit is capped at 200).
Parameters
| param | type | description |
|---|---|---|
| seed required | string | string[] | The seed company, or an array of seed companies. Each seed is a domain (e.g. "rippling.com"), a company name, or an org_id (from resolve_company). An unindexed domain is scraped from its homepage and embedded on the fly rather than rejected. |
| company_size | string | Optional firmographic narrowing to one employee-size bucket. Employee size bucket, exact values: "1 employee", "2-10 employees", "11-50 employees", "51-200 employees", "201-500 employees", "501-1,000 employees", "1,001-5,000 employees", "5,001-10,000 employees", "10,001+ employees". Compact forms like "51-200" or "1001-5000" are normalized. Optional. |
| industries | string[] | Optional firmographic narrowing: only return neighbors in these industries (fuzzy-resolved to the taxonomy, e.g. ["software", "financial services"]). |
| limit | integer | Neighbors per page (default 10, max 50). COST: each company returned costs 2.5 credits (plus a 0.1 base fee per call), so this limit sets the price of the call — ask for what you need. |
| locations | string[] | Optional firmographic narrowing: only return neighbors in these locations — city, region, or country (e.g. ["United States", "California"]). |
| offset | integer | Rows to skip for pagination (default 0). offset+limit is capped at 200 — similarity quality degrades deep in the neighborhood, so page depth is deliberately bounded. |
| rerank | boolean | Smart mode: an LLM business-model rerank of the top matches that reorders by what each company actually does and drops off-topic results — higher precision, but slower (~8s, plus a one-time model warm-up on the first call). Default false. |
Call it from your agent (MCP)
Connect once and lookalikes appears in your agent’s toolbox next to every other gtm.dev tool. Claude Code:
claude mcp add --transport http gtm https://mcp.gtm.dev \
--header "Authorization: Bearer YOUR_API_KEY"Under the hood, the client sends:
curl -X POST https://mcp.gtm.dev \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
"params":{"name":"lookalikes","arguments":{"seed":["rippling.com","gusto.com"],"industries":["software"],"rerank":true,"limit":25}}}'other clients (cursor, claude desktop, openai agents sdk, langchain): mcp setup
Call it over REST
curl -X POST https://api.gtm.dev/v1/tools/lookalikes \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"seed":["rippling.com","gusto.com"],"industries":["software"],"rerank":true,"limit":25}'Errors (401, 402, 429, 400) are listed in the REST reference. Calls that error are not billed — see pricing.