Two Make.com scenarios that run as one machine. Every day the first pulls new business-for-sale listings from the alert emails you already subscribe to, scrapes the full details, and builds a listing Rolodex, then matches every listing against each buyer client’s buy box. The second wakes up, reads the fresh matches, and drafts the outreach: a personalized email per hot match and a daily digest.
Listing alert emails land in Gmail; Claude extracts the listing links, and each one is checked against the data store, anything already known skips instantly. New listings get scraped in full by Firecrawl, normalized by Claude into structured fields (price, SDE, industry, location), and written to both the data store and the Rolodex sheet.
The router closes the loop, processed emails get relabeled, and fans into matching: the full buyer-client roster is pulled from Sheets, and Claude rates each new listing against every client’s criteria, producing a scored match row per pairing. The buy box does the filtering; the score does the ranking.
On its own schedule, scenario two reads the fresh match rows, skips anything already drafted, and splits: hot matches get a personalized Claude-written draft email in Gmail plus a note logged to Close CRM; the rest roll up into a Claude-written daily digest. Drafts wait for review, the digest sends itself.