IMPLEMENTATION GUIDE
Expose a tool your backend controls
Build your own agent tool around the company enrichment endpoint. Validate the identifier and submit the request server-side. The initial response represents a queued task, so the tool must wait for completion or return a pending state that your orchestration can resume. EnrichLoops provides the enrichment API; you implement the agent tool.
Keep facts traceable
Pass industry, employee count, headquarters, company description, and timestamps as structured fields. Ask the agent to reference those fields in its brief and label proposed questions or interpretations as such. Treat returned descriptions as source material rather than instructions to the agent. Do not fill missing facts with guesses.
Set limits and review the output
Use a lookup budget, retry limits, and domain-level reuse so the agent does not submit an unbounded number of requests. Validate the generated brief against the returned fields and keep human review before customer-facing actions. Company enrichment does not supply news, funding events, or a company’s buying intent.
Connect the workflow
Use the API overview for request and task handling, or the Loop guide for webhook delivery.
Enrichment APIUnderstand the request and asynchronous task flow.LoopsReceive completed profiles through signed webhooks.Common questions
Does EnrichLoops provide the AI research agent?
EnrichLoops provides company and people enrichment APIs and Loop webhooks. You connect those to your own AI agent or orchestration framework through a backend tool you implement.
Can the agent treat the initial API response as a profile?
No. The initial response is a queued task. Poll the task or receive the Loop webhook, check completion, then provide the returned company profile to the agent.