Agents do the investigation.
Your auditors decide.
Detection tells you where to look. The work that follows — records to collect, documents to read, exceptions to check — is what limits how many claims your team can cover. Our agents take on that investigation inside verification workflows taken from your own regulations, and put a prepared finding in front of your auditor.
line of eight stages, from selection to closure. Every page tells you its stage.
findings without their evidence. Each step shows what it read and the passage it rests on.
measures of an agent: stability across runs, agreement across models, agreement with your auditors.
from the day your extract lands to results on your own claims, in your own cloud.
How many suspicious claims can your team actually investigate?
Points. Investigates. Prepares. Leaves the decision to a person.
The console carries one claim from the first look to the last payment, and every person in the audit department finds their work in it. The stages that consume an auditor's week are done by agents. People keep the judgment.
Rules point
Sampling rules compare every facility with its peers, every period. A rule selects claims for review; it fines nothing. The head of audit steers coverage and the document budget, and sees the footprint of a threshold before choosing it.
Agents do the legwork
They request the documents, chase them, read them, and follow your audit procedure step by step: checks on the bill, checks with documents, exceptions. The result is a prepared finding with its recommendation, its evidence and its reasoning.
Your auditors decide
The auditor opens the file with the work done, inspects the sources, challenges what does not hold up, and confirms, changes or clears. Confirmed findings are grouped by facility into one notice that climbs your approval ladder.
A finding has to hold up when someone challenges it.
Running the same file again, or changing the model, should not change the answer. The workflows keep the audit bound to a defined procedure, whichever model performs the review, and the console measures it.
Workflows from your regulations
Each verification workflow is extracted from your regulations and audit manual, approved by your team, and enabled by a signed act before it runs on a single claim.
Same file, same answer
Stability across runs and agreement across models are measured per workflow and per step, continuously, and shown to the head of audit.
Measured against your auditors
Every decision your auditors take is also a review of the agent that prepared the claim. Where the agents agree with your team, and where they do not, is on one screen.
Your data stays in your account.
The console is deployed as a private tenant in your own AWS or Azure account, in your region. Nothing leaves it. Your extract, your rules, your users, your approval ladder.
Your extract
A de-identified claims extract for the period you choose. An adapter maps your fields onto the standard claim; a rule whose field is missing says so instead of guessing.
Your roles
Head of audit, chief auditor, auditor, finance, executive, and a portal for the facilities. Each sees the pages of their stage.
Your record
Every notice, appeal and payment stays on record; each facility's history feeds the next selection.
Your claims. Your cloud. Results in 30 days.
The thirty days start the day your extract lands.
Tenant and extract
We deploy the private tenant in your account and load your extract.
Your workflows
The workflows taken from your regulations are reviewed and enabled by your team.
Agents run, auditors decide
Every selected claim is investigated; your auditors review prepared findings on the console.
Results in hand
You hold the findings, the agreement figures and the record, and choose whether to continue.
Bring us your claims.
A private tenant in your own cloud, your extract, thirty days. Request it here, or find us on AWS Marketplace if your organisation buys there.