Quick: set up a data endpoint

Short recipe — one endpoint, no pipeline. For the full flow see Connect data.

Steps

1. In Integrations, start a new endpoint on the already-connected source system.
2. Name it for its consumers (crm.orders), namespace it by system (crm), and write the one-sentence description.
3. Set the role (source) and contract kind (query read for filtered pulls, stream read for ongoing sync).
4. Fill the target: the concrete table, path, or stream — at least one selector field, or the endpoint points nowhere.
5. Declare columns and the primary key as schema hints for mappers downstream.
6. Set owner team and freshness ("hourly"), and an honest trust rank — new doors start medium at best.
7. Validate and commit, then open the sample: right system, right objects, plausible freshness. If validation names a cause (wrong permissions, empty scope), fix that cause — not the whole setup — and revalidate.

Image: validation result naming the exact cause on a failed first attempt, then passing after the fix.

Image: endpoint form filled for crm.orders with validation passing.

Done means a workflow can name this door and read through it. Field detail in Data endpoints.