Skip to main content
Human guide: work with your agent → Resolve personal-email candidates for a small, authorized contact sample. Keep source evidence and unresolved rows. Finding an address does not establish permission to use it in a campaign.

​
Before you start

Prepare names and available person identifiers. The batch contract requires first_name and last_name per row. Its default columns also support linkedin_url, domain, and company_name. Use a correctly parsed CSV with a header. Follow file validation before running a provider-backed lookup.

​
1. Inspect the current batch contract

Compare your actual headers with the returned CSV contract. Use supported column mappings if they differ. Do not assume a LinkedIn-only file satisfies the required-name fields.

​
2. Run a small file

Prepare contacts-pilot.csv with representative records. Include an ambiguous or incomplete case so the pilot can expose the acceptance boundary.
This starts provider work and can incur usage. Save its run ID.

​
3. Inspect and export the same run

If several datasets are available, select the returned dataset path explicitly. Read source and attempt fields in the current output contract.
Reconcile every input to a returned result or an explicit unresolved state. Review the identity evidence and actual usage before expanding the file.
Do not describe consumer addresses as permanently current. A returned address also does not prove delivery, account ownership, or permitted downstream use. Keep the lookup and audience upload as separate approvals. Ask your agent to explain unsupported or missing results before retrying them.