The problem this solves
Names without addresses are the most common half-finished list: attendee exports, scraped speaker pages, the partner's spreadsheet of introductions. The people are right, the companies are right, and there is no way to reach them. Finding the addresses one at a time in a browser takes a minute each and produces guesses that bounce.
Two things make the completed file trustworthy. The finder should work from the company's actual domain and its known address pattern rather than a generic guess, and every address should be verified before it is trusted, with catch-all and risky results labelled rather than mixed in with the deliverable ones. The status column is what lets the file be used differently by a cautious sender and an aggressive one.
How the mission runs
- Read the file and resolve companies. The Enrichment Agent identifies the name and company columns, resolves each company to its domain, and normalizes names so that initials, nicknames and accents do not defeat the pattern match.
- Find a work address per person. The email finder searches for each person at the resolved domain, using the company's observed address pattern and any public trace of the address, and returns the best candidate with a confidence value.
- Verify and label every address. Verification checks each candidate and labels it deliverable, catch-all, risky or invalid. Invalid results trigger a second attempt with an alternative pattern before the row is given up.
- Return the completed CSV. The file comes back with email, status and confidence columns appended, original columns untouched, and a summary of the split between statuses and the rows where no address could be found.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
The original CSV with three new columns per row - work email, verification status and confidence - and a summary stating how many rows received a deliverable address, how many are catch-all or risky, and how many could not be found. The file is ready to filter by status and load into whichever tool sends the sequence.
Make it yours
- Add direct phone numbers as a second channel for the rows where no deliverable email exists.
- Load the deliverable rows straight into Lemlist, Instantly or Apollo as an unsent campaign audience.
- Save the completed file as list <name> so the addresses are available to every later mission.
- Ask for the company domain as an extra column so the file can be joined to other lists later.
Frequently asked questions
What does confidence mean beside a verified status?
Verification says whether the mailbox accepts mail; confidence says how sure the finder is that the mailbox belongs to this person. Both matter, and the file carries both.
Should I send to catch-all addresses?
A catch-all domain accepts everything, so the address cannot be confirmed. Many teams send to catch-alls with a lower bounce threshold; the label lets you make that choice per campaign.
What if a person has changed companies?
The finder checks the person's current employer where a profile is available and notes a mismatch with the company in your file rather than producing an address at the old employer.
How does it handle common names at large companies?
It uses the person's title, location and any profile URL in your file to pick the right individual, and lowers the confidence when several people share the name, so the confidence column tells you which rows to double-check.
Can I re-run only the rows that failed?
Yes. Attach the returned file and ask for a second pass on the empty rows; the agent tries alternative patterns and sources for those only, and the cost is limited to the rows retried.