Radar / Ideas / WireVerdict: the independent verdict…

WireVerdict: the independent verdict gate that stops BEC wire fraud before the money moves

daily ideamoderateJEV confidence 0.61JEV monetization 0.542026-10-16
OutcomeEvery vendor wire released with a fraud verdict an auditor can replay; BEC losses cut to near zero without slowing AP down.

The problem

Business email compromise cost $2.77B in reported losses in 2024 (FBI IC3), and the dominant pattern is vendor impersonation: a 'new bank details' email diverts a legitimate invoice payment because the invoice, amount and timing all look right. The AFP's 2025 survey found 79% of organizations were targeted by payments fraud and 63% faced BEC. The recommended control, callback verification of every bank-detail change, is manual and skipped at scale.

The idea

A vendor-neutral gate between the AP tool and the bank: every bank-detail change triggers automated out-of-band verification (callback to the pre-registered vendor number, micro-deposit confirmation, registry cross-checks), and every payment gets a calibrated fraud verdict from an open decision model with a signed, auditor-replayable evidence pack. Plugs into QuickBooks, Xero, Zoho Books or Ramp via API, so it works with whatever AP tool the company already uses.

Why now

Freehand Pay (Oct 15) just put autonomous AI agents inside AP execution, so agents that move money now need an independent check; calibrated decision models (autotrust JEV-35B, Decision 3.0, Oct 14-15) finally make per-payment fraud verdicts cheap and auditable; and Ramp itself is shipping AP agents (Oct 2026), concentrating more payment authority in software and making an independent verifier mandatory infrastructure.

What it combines

freehand-pay-ai-ap-teams-that-execute-payment-not-just-approval (autonomous payment execution, the thing that needs a check) + autotrust-jev-35b (calibrated approve/review/reject probabilities per payment) + optro-autonomous-grc-agents (the GRC evidence pack that makes each verdict auditor-ready). The mix matters because fraud detection without execution context is noise, and execution without an independent, replayable verdict is exactly what auditors reject.

MVP

Weekend MVP: a QuickBooks/Xero webhook listener that intercepts new vendor bank-detail changes, runs an automated callback plus micro-deposit check on one rail, and returns a Jev verdict (approve/review/reject with probability) plus a one-page evidence log. Deliberately skip: multi-rail execution, the full GRC evidence pack, and real-time payment interception.

How it makes money

Budget owner: Controller / CFO, from replacement budget: existing AP software spend plus fraud-loss reserves and cyber-insurance premiums. Pricing anchor: Tipalti at $99/month base or BILL at $45/user/month (https://ramp.com/blog/accounts-payable/best-accounts-payable-automation-software); price per verified bank change at ~$2-5, roughly 10x below enterprise vendor-validation engagements. Willingness-to-pay: 79% of organizations targeted by payments fraud in 2024 and 63% faced BEC per the 2025 AFP survey (https://www.cybersecurity-insiders.com/phony-bank-account-change-requests-how-to-detect-and-stop-aps-silent-killer/); FBI IC3 2024 reported $2.77B in BEC losses (https://www.infosecurity-magazine.com/news/fbi-staggering-lost-cybercrime-2024/).

The moat

The largest AP incumbents (Ramp at ~$13B, BILL) monetize platform lock-in and card interchange; a vendor-neutral verification layer that sits in front of any AP tool and any bank is a business-model conflict, since they would have to verify payments leaving their own rails. Pure email-security players (Abnormal, Proofpoint) inspect messages but cannot do out-of-band bank verification, an architecture they would have to build from scratch.

Distribution

B2B2C through accounting-platform app stores (QuickBooks, Xero, Zoho Books), where the AP clerks already work, plus partnerships with AP automation vendors that would rather integrate an independent verifier than build bank-verification infrastructure.

Why it wins

Ramp, BILL, Airbase and Tipalti fraud flags live inside their walled gardens and serve their own rails; Trustpair and nsKnox do vendor validation as enterprise sales-led engagements. WireVerdict is vendor-neutral (any AP tool, any bank), self-serve, priced per verified change rather than per seat, and every verdict ships with a replayable evidence pack for auditors.

Risks

Biggest risk is false positives freezing legitimate payments: AP teams will rip out anything that slows month-end close. De-risk by starting in monitor-only mode (verdicts logged, nothing blocked) and graduating to blocking only after precision is measured on the customer's real payment flow.

Build it with

Repo to start from

wireverdict-core: the verdict schema, the evidence-pack format, and bank-verification adapters; the hosted callback infrastructure stays closed.

Evidence

Get the week's best AI launches, plus 3 ideas worth building

One email every Saturday. Ranked by traction, not hype. Free.