Auto-enroll Airtable prospects into Lemlist with a custom icebreaker

Build me an agent workflow that automatically enrolls new Airtable prospects into a Lemlist campaign with a personalized icebreaker.

Trigger: poll my Airtable Prospects table for new records (new_record event). I'll wire up the base, table, and the Airtable connection when I install the workflow.

For each new row, do the following:

1. Read the prospect's first name, last name, company, role, any research notes I pasted into the row, and the Lemlist campaign slug from a field on the row (call it campaignSlug).

2. Call the Lemlist Enrich Data operation with the prospect's first name, last name, and company to find a verified email and LinkedIn URL. Enrichment is async: Lemlist returns an enr_ id, so wait for the job to finish before using the result (poll the enrichment status on a short interval, with a sensible timeout).

3. If enrichment fails to return a verified email, skip the row entirely. Do not enroll the lead and do not update Airtable. I do not want to burn campaign capacity or sender reputation on bad data.

4. If we got a verified email, generate a one-sentence personalized icebreaker. Ground it in the prospect's company, role, or the research notes from the row. No generic compliments like "love what you're building" or "impressed by your growth." It should sound like a human read their LinkedIn and the notes, and picked one specific hook.

5. Call the Lemlist Create Lead in Campaign operation against the campaignSlug from the row. Pass firstName, lastName, companyName, linkedinUrl, and the icebreaker as custom variables so my sequence template can reference {{icebreaker}}, {{firstName}}, etc.

6. Call the Airtable Update Record operation on the original row to set status to "Enrolled" and write the verified email into the email field. This keeps Airtable as the source of truth.

Use an agent workflow (not a deterministic graph) because the icebreaker quality and the skip-on-bad-enrichment decision both need judgement. Log each step so I can see why a row was skipped or enrolled when I open the run.