Use case · Fraud

A score says risky. It never says why.

Mysa screens every claim against fraud indicators and your own patterns, then surfaces the ones that warrant a look with the reasons attached — so your investigators spend time on the files that deserve it.

See how →
Screen a batch of your own claims · first pass on us
On this claim
1
Screen every claim
Run the fraud indicators over the file into Fraud Screen.
2
Weigh what fired
Test the indicators against your own patterns.
Research agent — Match the network and prior claims into Fraud Screen.matches.
3
Route it with the reasons
Attach the indicators and facts into Case, ready for the investigator.
Suggested by your decision graph
Hold and refer — this is a ring.
Source 3 unrelated claimants this quarter share one payout account; two more share the tow operator.
Why each claim looks clean on its own. The ring only shows across the book.
SIU deskLIVE
ScreenReviewSourcesReasoning
Automated · agentsYour call

Indicators ran on every claim, off the file’s own facts. What’s left is your call.

Screen readiness
Screened in full
18 indicators run
6 indicators fired
Late notice · prior similar loss · staged pattern
Network checked
2 links to a flagged ring
Next actions queued
Assign SIU · request EUO
Same repairer and passenger as a claim denied 4 months ago. Worth a look.
This pattern turned out fraudulent in 12% of matched claims.
Ready to review

The score told you it was risky. Not why.

A black-box number flags a claim and leaves the investigator to rebuild the case from scratch — or wave it through. The reasons are the value.

FR-2201screened · clean—
FR-22026 indicators firedneeds eyes
FR-2203screened · clean—
FR-2204screened · clean—
FR-2205network match · flagged ringneeds eyes
FR-2206screened · clean—
How it works

Every signal on the claim, wired to the reason it fired.

Not a number. The indicators, the network, and the patterns on your own book — connected to the call.

Every claim, not a sample

Indicators run on the whole book automatically — nothing skips screening for lack of time.

Screen
FR-2201 · screened
FR-2202 · screened
FR-2203 · screened
Full book screened

The reasons, not just a score

Each flag arrives with the indicators that fired and the facts behind them, ready to investigate.

Explain
Late notice
Prior similar loss
Staged pattern
6 reasons attached

Tuned to your own book

Flags reflect the patterns on your own claims — which cuts the false positives that bury investigators.

Tune
Generic model · 40 flags
Your patterns · 6 flags
Fewer, truer flags

Starts with the indicators you know. Sharper every file.

The instinct for a claim that doesn’t smell right is hard-won. Mysa keeps it.

Day one Your indicators, encoded
The signals your team already watches, run on every claim from the first day.
Every call Their instinct, captured
Every claim your team clears or pursues teaches it what fraud looks like on your book.
Over time Flags like your best investigator
It surfaces the right files with the right reasons — and it stays when they leave.
Day oneEvery callOver time
On your terms

Your files. Your call.

Sovereign by design

Your cloud, on-prem, or your VPC. Claims never leave your environment.

Your team keeps the call

Mysa flags and explains; the decision to investigate or deny stays with a human.

Defensible by construction

Every flag cites the indicators and facts behind it, logged and explainable.

Common questions

How does Mysa detect claims fraud?

It screens every claim against fraud indicators and the patterns on your own closed claims, then surfaces the ones that warrant a look with the specific reasons and facts attached. Claims leakage detection

Does Mysa reject claims for fraud on its own?

No. It flags and explains; the decision to investigate or deny stays with your team. It is a human-in-the-loop system by design. Human in the loop

How does it reduce false positives?

By learning from the patterns on your own book rather than a generic model, the flags reflect what is actually unusual for your claims, which cuts the noise that buries investigators.