4,535 rows in. 2,953 real people out.
The lead engine behind our outbound, built to white-label. It takes any export, maps the columns, removes duplicates, researches every company, writes the opener and puts it all in one dashboard.
Lead Dashboard
Prospect research and outreach lines, one row per person.
Clean first. The AI comes last.
The whole idea in 60 seconds, one product decision at a time.
Every source speaks a different dialect.
Apollo, Google Maps and directory scrapes all name their columns differently. A normaliser maps them to one schema before anything else runs.
Company Name for EmailsCompanytitleCompanyE-mailEmailemails[0]Emailcompany domainWebsitePerson Linkedin UrlLinkedIncity, stateCitySame person, three lists, one row.
Emails are trimmed and lowercased into a match key. Matches merge, role inboxes get flagged, and domains are grouped to catch guessed addresses.
Dana@NorthwindDental.CA dana@northwinddental.camarco.ruiz@peakfit.cominfo@lakeshorepizza.comDANA@northwinddental.ca leo.tran@peakfit.comdana@northwinddental.canew persondana@northwinddental.ca1 sourcemarco.ruiz@peakfit.com1 sourceinfo@lakeshorepizza.comROLE INBOX · RISKYleo.tran@peakfit.com1 sourceSeven stages, one click, five minutes.
A logged run of 100 leads: 85 homepages read, 82 summarised by Claude Haiku, 77 specific lines, 23 fallbacks, $0.065 in model cost.
An observation, then a bridge from the same pain.
Each line is sorted into a pain family by free rules, and the bridge comes from that family's approved bank. The line and the pivot always agree.
A tool a non-coder can run.
No terminals. Every job has a web screen, and an assistant can do file work for you with guardrails.
Sent history on the left, new export on the right.
Drop the files, run it, get a clean CSV that skips anyone already contacted.
Ask in plain English.
Claude calls file tools to count, compare, dedupe and rename lists.
Remove everyone we already emailed from the new clinic list.
dedupe_csvs(existing=2, new=1)Removed 41 already contacted and 3 internal duplicates. 286 ready to send.
Nothing is ever hard-deleted.
File work is sandboxed to the project, and "delete" moves the file to a dated trash folder.
Progress saved every 10 leads.
A dropped connection mid-run resumes where it stopped. Restart or jump to any stage with one flag.
CSV files, no database.
Python, pandas and Flask on a laptop. The only paid call is the model, a few cents per hundred leads.
Want a lead engine that runs under your brand?
We build the pipeline, the dedupe and the dashboard around your sources and your offer, then hand it over or run it for you. Start with a free 20-minute call.