Follow one record through the machine.
With FiveFacts AI, a shop loads a CSV and headers map automatically. Every record is verified at intake, because guessing a merchant's industry from the business name is right only about 40% of the time. Voice and SMS then work the book as one thread inside each state's calling window, qualify the five facts, and hand the closer a warm merchant with the application and bank-connect link already sent.
The homepage gives you the outline. This page is what's underneath it: what actually happens to a piece of aged paper between the CSV and your closer's headset, in the order it happens.
From file to warm handoff, in order.
Load the book.
A CSV or a spreadsheet, exported from whatever you run today. Headers map automatically, whatever your last few systems did to the column names, and records dedupe at intake. No CRM surgery and no six-week implementation, because there is nothing to implement. The heaviest lift on your side is finding the export button.
Every record gets grounded before anyone dials it.
Between upload and first dial, the machine verifies who it is about to talk to. The business gets looked up, the industry gets confirmed against what the business actually does, and the verified vertical is stored on the record once, then reused on every touch after. A record that can't be verified opens neutral instead of confidently wrong. The 40% section below is why.
Voice and SMS work the merchant as one thread.
One conversation with one memory: a missed call becomes a text in the same thread. Dials fire only inside the merchant's state calling window and never in quiet hours. Retries run in rounds across every number on file instead of hammering one line all afternoon, and a brush-off gets a scheduled comeback instead of a shrug.
Five facts, asked in conversation, logged every time.
Monthly revenue. Time in business. Amount seeking. Open positions. State. Asked the way a good opener asks, inside a conversation the merchant is actually having, not read off a form. Every answer lands on the record. A stall on the last fact is not a qualified merchant; five facts is. The grading never loosens at the end of the month.
The handoff happens while the merchant is still warm.
The moment a merchant qualifies, the application link and bank-connect link go out by text, in the same thread, while the conversation is still open. Your closer gets notified with everything attached: the five facts, the contact history, the whole exchange. No re-pitching, no "who told you that." You watch it all in a read-only live console, and a branded weekly report goes out that you can forward without touching.
The 40% problem.
Guessing a merchant's industry from the business name is right about 40% of the time on hard names. Measured, not estimated. Here is what that number does to an aged book, and what the machine does instead.
Open on a guess
"Marra & Sons" could be a paving crew, a fish market, or a law office. Open with the wrong one and the merchant doesn't correct you; he files you with every mail-merge that never knew him and hangs up. An aged lead gives you one re-opening. A wrong industry guess spends it on the spot, and the record comes out colder than it went in.
Open on a verified fact
Every record is grounded at intake: business looked up, vertical confirmed, stored once, reused on every round after. That lands at roughly 86% verified grounding before the first dial. When Veronica opens on what the merchant actually does, he hears somebody who knows his file. And when a record can't be verified, she opens neutral, because generic beats wrong.
The gates every touch clears.
Every touch, first dial of the day or last text of the round, clears the same checks.
The responsibility split: consent basis and list origin belong to you. From the moment data hits the machine, calling windows, opt-outs, gates, and the audit trail are enforced in code by us.
Compliance machinery, not legal advice.
What going live looks like.
Your number, your name
Caller ID, every text, the application link: all yours. Veronica introduces herself as your shop's opener because, on your deployment, that is exactly what she is. Merchants never hear our name.
One deployment per shop
Your book runs on a machine that runs only your shop. Nobody else's records sit next to yours, and yours teach nobody else's opener.
Days, not quarters
Onboarding is the file, your brand assets, and a working session on script posture: how your shop talks, what it offers, where the lines are. Then the console lights up and the first records get worked.
The copy keeps auditioning for its job.
Every message variant runs against a control, and reply rates are measured per variant, continuously. Nobody's opinion of what sounds good survives contact with the measurement, including ours. The receipts show up in your weekly report.
The machine doesn't have a good day or a bad day. It has the same day, on every record, inside every window, until the book is worked.
See it work a record you own.
Apply for the demo and play the merchant: stall her, lowball your revenue, brush her off. Then load the book, or a slice of it, and watch a live run on your own paper instead of anybody's brochure.