GTM data for agents. pipe0 tools for the Vercel AI SDK: find people and enrich them with verified work emails, phone numbers, profiles, and firmographics from dozens of data providers.
Each tool runs a pipe0 waterfall: providers are tried in order until one finds the value, so your agent gets the best available coverage from one call and one API key.
| Tool | What it does |
|---|---|
findPeople |
Search for prospects by job title, employer, seniority, job function, company size, and location. |
enrichPerson |
Find a person's work email, mobile number, and profile from a LinkedIn URL, an email, or name + company domain. |
enrichCompany |
Get firmographics for a company domain: description, industry, headcount, revenue, founding year. |
npm install @pipe0/ai-sdkGet an API key at app.pipe0.com and set it as PIPE0_API_KEY. New accounts include free credits.
import { generateText, isStepCount } from "ai";
import { pipe0Tools } from "@pipe0/ai-sdk";
const { text } = await generateText({
model: "openai/gpt-5-mini",
prompt: "Find the CTO of Linear and get their work email.",
tools: pipe0Tools(),
stopWhen: isStepCount(10),
});
console.log(text);Or pick individual tools:
import { enrichPerson, findPeople } from "@pipe0/ai-sdk";
const tools = {
findPeople: findPeople(),
enrichPerson: enrichPerson(),
};Every tool (and pipe0Tools) accepts the same options:
pipe0Tools({
apiKey: "...", // defaults to process.env.PIPE0_API_KEY
environment: "sandbox", // "production" (default) or "sandbox": free placeholder data for development
});You can also pass a configured client from @pipe0/client.
Every call spends credits in production. In agents where users trigger the calls, ask for approval with the AI SDK's toolApproval:
const result = await generateText({
model: "openai/gpt-5-mini",
prompt: "Get the work email of the CTO of Linear.",
tools: pipe0Tools(),
toolApproval: {
enrichPerson: "user-approval",
},
stopWhen: isStepCount(10),
});On AI SDK 5 and 6, which have no toolApproval, pass needsApproval: true to the tools instead. It is deprecated on AI SDK 7.
Tool calls run in production by default and spend pipe0 credits: roughly 0.1 credits per findPeople result and from 0.5 credits per value found by enrichPerson. Use environment: "sandbox" while building. See pricing.
Run the tools on your server and keep PIPE0_API_KEY out of client bundles. Anyone with the key can spend your credits.
findPeople and enrichPerson return personal contact data. Make sure your use complies with the privacy and marketing rules that apply to you (for example GDPR or CAN-SPAM), and only show results to users allowed to see them.
pipe0 covers much more than search and enrichment: buying signals (job changes, funding, new hires), routing leads to HubSpot, Salesforce, Attio, or Slack, and scheduled lead lists in pipe0 sheets. Use the REST API and @pipe0/client for the full catalog, or connect the pipe0 MCP server to an agent.
MIT