WEB DATA & EXTRACTION · FOR AI AGENTS
Crunchbase API
Retrieve licensed company, funding and relationship data for research workflows.
Crunchbase’s read-only REST API lets research agents look up organizations, search licensed collections and retrieve related records. Field availability follows the provisioned data package, making it useful for structured company research and enrichment.
Your integration checklist
- Capabilities
- What you need from your owner
- Arrange REST API access through sales and obtain a provisioned key. Confirm the licensed fields, intended use and sharing rights before integrating. A separate MCP connection does not supply REST credentials.
- Authentication
- Send the key in X-cb-user-key to https://api.crunchbase.com/v4/data/. Keep the key in a secret store rather than in a request URL. Use the reference matching the licensed package.
- Your first task
- Look up one known organization, validate its identifier and domain, then request the smallest licensed field set needed for a research brief.
- Billing & payment
- REST access is provisioned through a commercial agreement; fields and collections depend on the package. MCP access and quotas are separate. The legacy Basic API is no longer offered to new users.
- Agent payment support
- Not verified. Programmatic usage does not establish autonomous checkout support.
- Operational limits
- Enterprise REST keys have separate request-rate and concurrency limits: 200 requests per minute and three in progress. Missing fields can reflect access or query issues. Company data and predictions are not guaranteed current facts or investment outcomes.
ORIGINAL AI AGENT STORE INTEGRATION PLAYBOOK
Keep entity identity and missing-data reasons visible in enrichment
Editorial recommendation · Reviewed 2026-11-08
When to choose this service
Choose Crunchbase for licensed structured company relationships and funding research. Direct company filings or announcements may suffice for a small number of known organizations.
The maintained company and relationship dataset is an external research resource. An agent can organize records but cannot create reliable unobserved funding history or licensed fields from reasoning alone.
Suggested workflow
- Match the organization using its identifier and domain, then request only the fields required for the output.
- Share rate and concurrency limits across workers. Retain a pagination cursor only after validating and storing the corresponding response.
- Separate returned facts, missing fields and editorial conclusions. Preserve provenance and check whether the license permits the intended export.
Acceptance checks before real work
- A same-name organization may be a different legal entity.
- A missing field does not mean the underlying fact is zero.
- Partial pagination is an incomplete dataset even when earlier pages succeeded.
If the workflow fails
On a rate-limit response, honor Retry-After where supplied and retry through the shared queue. For missing fields, check spelling and licensed access before repeating a large search.