The platform
A verification engine, not another risk score.
Vericlaims runs a fleet of specialist agents over the evidence attached to a return or a claim, corroborates it against your systems of record, and returns a verdict with its reasoning attached — through one API and one reviewer console.
A probability that something is wrong.
Rules and models rank suspicion from transaction patterns. Useful for triage, but it cannot tell you whether the parcel was empty or the invoice was real — so the case still lands on a human desk, and the customer still waits.
A finding about what actually happened.
Agents interrogate the evidence itself and reconcile it against your records. The output is a verdict — approve, refer or decline — with the specific signals, weights and rationale that produced it, ready for a reviewer, a customer or a regulator.
Verification agents
Specialists, not one model doing everything.
Each agent owns a class of evidence and a class of question. They run in parallel, hand their findings to adjudication, and every one of them shows its working.
Provenance agent
Establishes where an image or video came from: capture metadata, re-encode history, generative-model fingerprints and reverse-match against known stock and prior submissions.
Document forensics agent
Reads invoices, receipts, estimates and reports — checks fonts, layout lineage, arithmetic, issuer records and template reuse across your book.
Reconciliation agent
Matches the story to the system of record: order lines, payment events, courier scans, policy terms, coverage windows and prior claims.
Network agent
Links identities, addresses, devices, bank details and repairers to surface organised rings and repeat actors without pooling raw personal data.
Policy agent
Applies your written policy and regulatory constraints to the evidence, so decisions stay inside the rules you already operate under.
Adjudication agent
Weighs the signals, resolves conflicts, sets confidence and writes the rationale that a reviewer, customer or regulator will read.
Evidence coverage
Everything a case actually arrives with.
Photos
Capture metadata, EXIF, re-encode history, generative fingerprints
Video
Frame continuity, splice detection, live-capture attestation
Receipts & invoices
Issuer records, template reuse, arithmetic, layout lineage
Reports & estimates
Repair estimates, medical paperwork, loss reports
Device & session
Device integrity, location plausibility, session lineage
Records & graph
Orders, payments, courier scans, policies, prior claims
Deployment
Earn the automation before you switch it on.
Most teams start in shadow mode, compare verdicts against their own decisions, then move line by line. You choose the thresholds, and you can move back at any time.
Shadow
Runs alongside your current process on live or historical cases. Produces verdicts you can compare against what your team decided — no customer impact.
Assist
Verdicts and dossiers land in your reviewer queue. Humans still decide, but the collection and checking work is already done.
Autonomous
Clean cases clear straight through against thresholds you set; only referrals reach a human. Every automated decision stays fully auditable.
A reviewer console your investigators will actually use.
Referrals arrive as a prepared dossier: the evidence, every check that ran, what each one found, and what remains unresolved. Reviewers confirm, override or request more evidence — and every action is logged against the case.
- Queues and thresholds per line of business
- Side-by-side evidence with the anomalies highlighted
- One-click override with a required reason, captured for audit
- Exportable case file for disputes, chargebacks and litigation
See it run on evidence that looks like yours.
A 30-minute walkthrough: we take a case type from your book, run the verification live, and show you the rationale it produces.