The Payment Request That Wasn't Real: Verify First
The most dangerous moment in home admin is a plausible, urgent request that rides on top of a real transaction. Here is the sequence that catches it before your money leaves.
The short answer
To verify a contractor, adjuster, or payment request is real, stop and confirm it out-of-band: call the party at a number you already had, not the one in the message. Match every detail against your existing file, treat any last-minute change to payment instructions as fraud until proven otherwise, and never act on urgency alone.
The request that looked exactly right
A homeowner three weeks into a roof claim gets an email from her adjuster. Same name, same claim number, same signature block. It says the carrier is expediting her payment and needs her to confirm bank details through a new portal by end of day. She almost clicks.
The tell was small: the reply-to address was one character off, and her real adjuster had never once mentioned a portal. She called the number on her original claim letter, not the one in the email. The adjuster had sent nothing. The email was a spoof built on top of a claim that was completely real.
This is the pattern that costs people the most. Not the obvious Nigerian-prince nonsense, but a legitimate transaction that a scammer has quietly climbed onto. The claim is real. The contractor is real. The closing is real. Only the payment instruction is fake, and it arrives at exactly the moment you are expecting one.
Why urgency plus legitimacy is the danger zone
The core mechanism
Fraud succeeds when a request is both plausible and urgent. Plausibility lowers your suspicion; urgency removes your pause. Scammers time impersonation to real events, a claim payout, a closing, a contractor's final invoice, so the fake request feels like the thing you were already waiting for.
The money in home transactions is large, one-time, and moves by wire or ACH, which is hard to claw back. That combination is why real estate and home services are prime targets.
According to the FBI's Internet Crime Complaint Center (IC3), business email compromise and real estate wire fraud account for billions in reported losses annually, and the median victim is not careless. They are busy, mid-transaction, and acting on a message that matched their expectations.
The uncomfortable truth: your own competence works against you here. You are used to handling home admin fast. A scammer's entire play is to make the fake request feel like one more thing to knock out before lunch.
The 5-step verify-before-you-act sequence
This is the exact sequence an AI verification layer runs on an inbound request, and the one you can run yourself. The agent's job is not to be smarter than the scammer. It is to refuse to act on anything it cannot authenticate, which buys the pause the scammer is counting on you skipping.
- 01
Match the request against the known party
Pull the request against your existing file. Does the sender address, phone, name, and claim or job number match the party you already dealt with, character for character? A one-letter domain change, a new phone number, or a first-name-only signature where you always got a full one is a hard stop. The agent flags mismatches automatically; you confirm the pattern.
- 02
Confirm out-of-band on a number you already had
Never call the number in the message or reply to the email. Call the contractor, adjuster, or title company on the number from your signed contract, your original claim letter, or their official website. If the person on that line did not send the request, you are done. This single step defeats most impersonation.
- 03
Flag anomalies before acting
Any change to payment instructions is guilty until proven innocent. New bank account, new portal, a 'we switched processors' note, a wire when you expected a check, a rush deadline that did not exist yesterday. These are the anomalies that ride on real transactions. The agent surfaces them; it does not smooth them over.
- 04
Refuse if it cannot be authenticated
If the request cannot be matched to a known party and confirmed out-of-band, the answer is no. Not 'later,' not 'probably fine.' A verification agent's default is stop, not send. It will not forward money, update payment details, or sign anything it cannot authenticate, even if the message is angry or urgent.
- 05
Escalate to a human with the evidence
When something is genuinely uncertain, the agent escalates to you with the specific tells laid out: the mismatched address, the changed account, the timeline that does not fit. You make the call. The machine catches the pattern; the human owns the decision. That division of labor is the whole point.
Red-flag reference: what a fake request looks like
| Signal | Usually legitimate | Likely fraud |
|---|---|---|
| Payment instructions | Consistent with your contract or prior invoices | Changed at the last minute or introduced by message |
| Sender address | Exact match to prior correspondence | One character off, or a new free-email domain |
| Urgency | Reasonable timeline, no threats | 'Today only,' 'wire now,' penalty language |
| Confirmation | Happy to verify by phone on your known number | Resists a callback, gives only a new number |
| Channel | Same channel you have been using | Sudden switch to text, portal, or new email |
| Payment method | Check or established ACH | Wire, gift card, crypto, or 'new processor' |
No single row proves fraud. But two or more together, especially a changed payment instruction plus a new deadline, should stop you cold. The scammer needs you to treat each anomaly in isolation. A good verification layer, human or agent, reads them together.
Why the agent's default is stop, not send
An honest AI agent handling your home admin is built to be conservative on money and identity. It will happily draft, organize, remind, and compare. It will not move funds or change payment details on its own judgment, because the cost of a false 'send' is catastrophic and the cost of a false 'stop' is a two-minute phone call.
This is where people misunderstand what AI should do here. The value is not autonomy. The value is a tireless second set of eyes that never gets rushed, never assumes, and never lets a plausible message skip the callback. At One Home Agent, the insurance agent (Gloria) and vendor agent (Vinny) treat any payment or identity change as an escalation event, not a task to complete.
“The best thing an agent can do with an urgent payment request is refuse to be hurried. It flags the tells, it makes the call, and if it still cannot authenticate the request, it stops and hands you the evidence. The scammer's entire plan depends on nobody pausing. We build in the pause.”
Todd Paton, Partner, One Home Agent
Checklist
0/7Run this before you send a dollar or change any payment detail
Bottom line
The most expensive mistake in home admin is acting fast on a request that looked right. Slow down at exactly the moment you are urged to speed up. Match, confirm out-of-band, flag the anomaly, and refuse anything you cannot authenticate. A verification layer, human or agent, exists to protect that pause.
Get a verification layer on your home admin
A second set of eyes that never gets rushed
One Home Agent adds a verification layer to your bills, claims, and contractor payments: it flags spoofed requests, out-of-band changes, and last-minute payment switches, and it refuses to act on anything it cannot authenticate. See how it works.
Talk to usFrequently asked questions
Confirm it on a phone number from your signed contract or the contractor's official website, never the number in the message. Check that the amount, job number, and payment method match prior invoices. Any new bank account or last-minute wire request should be treated as fraud until you verify it by voice.
Sources & further reading