Can Trestl flag underpayments?

Yes. Amount checks compare the paid amount with the amount the plan schedule produces and flag differences in either direction.

Do we have to use the Trestl review workspace?

No. Findings can be reviewed in the Trestl workspace or returned to your adjudication system or TPA through the exchange, depending on how you want your analysts to work.

How are AI-assisted checks governed?

They run on Google Cloud Vertex AI inside our Google Cloud business associate agreement, return structured results, and only ever raise items for human review. If the AI service is unavailable, the check does not fire rather than guessing.

Can checks be tuned for our plans?

Yes. Each check is configured from your plan terms, and settings such as the waiting period and the window for spotting duplicate claims can be set for your environment. Each check is tested with your team, on sample or historical claims or in a shadow review, before you rely on its findings for live claims.

Platform

From claim intake to a recorded decision, in one place

Trestl runs configured checks on each claim line before payment, routes findings to the right analyst with the evidence attached, and exchanges data with the systems you already run.

See it on synthetic data

The claim path

Receive, check, review, return

Step 1
Receive
Claims arrive the way you already move them: SFTP, API polling, cloud storage, or file upload.
Step 2
Check
Each claim line runs through the checks configured for its product and plan, before payment.
Step 3
Review
Findings land in an analyst queue with the rule, the expected amount, and the evidence.
Step 4
Return
Decisions flow back to your adjudication system or TPA through the same channel.

Prepayment checks

Plan logic for fixed-benefit lines

Checks are configured per client and per plan. Each product line has its own set, maintained as named rules with tests, so a finding can always be traced to the provision behind it.

Hospital Indemnity (HIP)

Benefit amounts · Employer Designated Facility multipliers · Waiting periods · Overlapping stays · Plan-year utilization limits · Duplicate claims

Hospital Indemnity checks

Accident (ACC)

Benefit amounts · Per-accident and plan-year limits · Duplicate incidents · Duplicate claims

Accident checks

Critical Illness (CI)

Covered diagnosis · Recurrence separation · Subsequent-diagnosis rates · Relationship percentages · Benefit amounts · Clinical document review

Critical Illness checks

Analyst review

Every finding shows its work

Findings land in a queue organized by product and assignment. Analysts see what fired and why, decide, and move on; managers see workload across the team.

The rule that fired
A named edit code tied to the plan provision it enforces, not an opaque score.
Expected versus paid
For amount checks, the amount the plan schedule produces next to the amount on the claim.
The evidence
Claim history, plan configuration, and, where relevant, the documents an AI-assisted check compared.
The decision trail
Who reviewed it, what they decided, and when, written to the audit ledger.

Role-based access

Analysts, managers, and client users each see their own scope. Client users only ever see their organization’s claims.

Assignment and workload

Claims are distributed to analysts by product, with queue and workload views for managers.

AI assists, people decide

AI-assisted checks compare accident reports and clinical documents and raise items for review. They never deny a claim.

Complete audit trail

Views of claims data and every decision are recorded in an append-only, hash-chained ledger.

Exchange

Connect the way you already move claims

Each client environment gets its own connectors for claims, members, plan configuration, and supporting documents in, and results out.

SFTP

Scheduled polling of your SFTP location; results written back to an agreed folder.

API polling

Authenticated polling of a partner API, with JSON results returned.

Cloud storage

A Google Cloud Storage exchange with inbound, outbound, and archive areas.

Ingest API

An authenticated endpoint that accepts a CSV claims batch with a key issued for your organization.

The route, file contract, schedule, and acceptance tests are agreed for your environment during implementation and validated with a representative file before any live data is sent.

Postpay investigations

A workspace for claims that were already paid

Some questions only surface after payment. Investigators work cases with the same claim history and audit trail, with workload management and recovery tracking.

Case workload

Open investigations organized by status and owner, per client.

Historical diagnostics

A bounded review of paid claims under an agreed scope, reported with the checks completed and any data gaps.

Recovery tracking

Outcomes recorded separately: a finding, an accepted correction, and a confirmed payment outcome.

Same controls

The tenant isolation, access control, and audit trail that protect prepayment review.

Frequently asked questions

Can Trestl flag underpayments?: Yes. Amount checks compare the paid amount with the amount the plan schedule produces and flag differences in either direction.
Do we have to use the Trestl review workspace?: No. Findings can be reviewed in the Trestl workspace or returned to your adjudication system or TPA through the exchange, depending on how you want your analysts to work.
How are AI-assisted checks governed?: They run on Google Cloud Vertex AI inside our Google Cloud business associate agreement, return structured results, and only ever raise items for human review. If the AI service is unavailable, the check does not fire rather than guessing.
Can checks be tuned for our plans?: Yes. Each check is configured from your plan terms, and settings such as the waiting period and the window for spotting duplicate claims can be set for your environment. Each check is tested with your team, on sample or historical claims or in a shadow review, before you rely on its findings for live claims.

See your claims through a prepayment lens

Start with a conversation about your book, the claim decisions that need attention, and who owns them. We will show the platform on synthetic data and agree whether a scoped review makes sense.

Request a demo Security & trust