HealthAura: Claims Audit for Health Insurers and TPAs
Every claim checked before payment and after. Every finding cites the rule it used, so it can be put to a provider or a regulator.
Run a claim through HealthAura
Pick a scenario, choose before or after payment, and watch the engine check each line, cite its rule and hand your team a decision.
- Reads the claim line by lineWaiting
- Coding and billing rules (DHA, DoH, MOHAP)Waiting
- Clinical indication against the diagnosesWaiting
- Eligibility and approvalsWaiting
- Paid-claims historyWaiting
Every claim line checked, before payment and after
Coding and billing
Unbundling, duplicates, price above tariff, approvals, age and gender
Clinical indication
Every service checked against the diagnoses on the claim
Regulator coding guidelines
Today: ICD-10-CM validation and the 2018 edition for Sharjah
Next: ICD-10-CM 2024 when DoH moves
ADOC readiness
Abu Dhabi claims read in current codes and in ADOC, side by side, before the switch.
By mid-November 2026: shadow billing
One agent for each job your team does by hand
Three AI agents check every claim and hand you a worklist. Each runs over the whole file on its own, and every finding cites the rule it used.
Eligibility
Reads the eligibility log
- Never checked, or checked late
- Billed after a “not eligible” answer
- Bulk night or after-hours checks
Approvals
Reads the policy table of benefits and the provider manual
- No approval at all
- Approval not matching the provider manual or table of benefits
- Releases by override, listed by name
- Records who processed each approval
Paid claims
Reads every claim and what NAS paid
- Paid, but the rules reject it
- Paid above the allowed amount
- Paid twice: same visit date, diagnosis and doctor
Demo day: 4,872 generated claims, built so every finding type shows; says nothing about NAS’s rates.
One goal, its own checks, every step on the record
Goal: is this reimbursement genuine?
Reads, checks, links, ranks and drafts.
Holds a claim, contacts a facility, sends a letter, starts a recovery.
Pays, denies or recovers. Those stay with your team.
AI reads documents, links records, ranks and drafts. It never decides what NAS pays or recovers.
Six agentic AI capabilities, grouped by what your team needs
Check every claim
- Coding, billing and clinical rules
- DHA, DoH and MOHAP rule sets, ADOC
- Eligibility and approval checks
- Each finding with its denial code
Recover overpayments
- Recoupment list from remittances
- Audit sample for clinical review
- Provider letters, offsets and disputes
- Repeat providers to pre-payment review
Find fraud, waste and abuse
- Patterns against peers, week by week
- Provider and pharmacy shopping
- Rings across members and providers
- Alerts ranked by money and evidence
Trust every document
- Read across approval, claim and audit
- Genuine, true, part of a pattern?
- AI-made fakes screened
- Visits confirmed with the facility
Stop it before payment
- Reimbursements checked before payout
- Approvals checked against history
- Watched providers held for review
- Same agents in claims processing
Defend every finding
- Rule, version and evidence on each line
- An auditor approves every action
- Full audit trail for regulators
- Data and models stay in the UAE
A recoupment list the day the remittance lands
Recover what was paid wrong
From the first run. Rule, adjudicator and payment reference on every row.
Every finding traceable
Rule findings go to recovery. Peer patterns go to the clinical sample, never a demand.
Track each line to closed
Actions logged under the auditor’s name. Repeat providers can move to pre-payment review.
A first run on one month NAS has already audited, to see what it adds
Demo day: 4,872 generated claims from 52 providers. Amounts are what the rules attribute, not a legal claim on a provider.
See HealthAura on a month you have already audited.