The decision layer for claims.
A claim moves as fast as its slowest decision. Medley makes most of them on arrival, at the maturity level you set. You settle faster, leak less, and can show a regulator exactly why.
species: dog
vuln: false
est: €680
Priced on claims handled, not seats.
Every tier includes the full decision trail, EU residency and dedicated onboarding. No hidden enrichment costs.
One corridor. Ninety days.
Shadow-mode on your live claims. Your ops run as normal, Medley runs in parallel and shows you exactly what it would have decided.
- ✓One corridor, shadow deployment
- ✓Full Synthesise → Enrich → Decide pipeline
- ✓Up to 10,000 claims
- ✓Forward-deployed engineer, embedded
- ✓DecisionRecord decision trail
- ✓ClaimOps dashboard
- ✓End-of-pilot business case
Carrier-grade. Per claim.
Full production deployment. Multiple corridors, carrier-tunable thresholds, every maturity level available as each gate earns it.
- ✓Unlimited corridors
- ✓Five maturity levels (L1–L5)
- ✓Learning Loop · threshold recommendations
- ✓REST, MCP and webhook integration
- ✓99.9% SLA · EU residency guaranteed
- ✓ISO 27001 · SOC 2 Type II audit underway
- ✓Customer success manager
* standard corridor · rises with corridor complexity and maturity level. Volume discounts from 5,000 claims / month.
Group carriers & platforms.
Volume pricing, dedicated infrastructure, custom integrations and co-built specialty corridors.
- ✓Everything in Production
- ✓Dedicated tenant infrastructure
- ✓Custom corridor development
- ✓Embedded and white-label options
- ✓Executive business reviews
- ✓Custom SLAs and DPA
Run the numbers on your own book.
The saving comes from claims that never reach an adjuster. Set your monthly volume and what a claim costs you today to see what that is worth over a year.
- 78
- auto-resolved — no adjuster ever opens them
- 22
- routed to an adjuster with the context pack already built
Out of every 100 claims, at L3 maturity.
* Illustrative only. Assumes L3 maturity (~78% auto-resolve), a mid-range corridor mix and EU claim volumes. Actual savings depend on maturity level, claim complexity and corridor configuration, modelled precisely in a discovery call.
Four production corridors. Live today.
Configured for the corridors carriers actually write. Carrier-tunable, replay-tested, each with its own SLA gates.
A governed decision layer — not another point tool.
Every claim becomes a signed, replayable decision.
Not a workflow. Not a chatbot. A governed decision layer that reads policy, weighs evidence and outputs an auditable verdict in milliseconds.
Your maturity level. Your rules.
Start at rules. Earn autonomy one gate at a time. Each level graduates on its own accuracy, drift and fairness metrics, and you sign the promotion.
Every decision signed & replayable.
DecisionRecord on every verdict. Model version locked. Prompt SHA-pinned. Replay any claim against any version.
Atomic decisions
Atomic auto-resolve outpaces claim STP by 16pp — because most claims have one or two hard decisions, not five. Adjusters work the 7%, not the 78%.
One decision layer, every book it runs.
Same product, different thresholds. Each carrier sets its own corridors, its own confidence floors and its own authority caps. What does not change is that every decision is recorded, replayable and owned by someone.
Synthesise. Enrich. Decide.
Three pillars, one governed decision layer. Communications, data integrations and the Learning Loop sit underneath, wrapping all three.
Synthesise
Scattered inbound becomes one structured claim record. Key fields extracted, vulnerability and injection flags raised before any decision logic runs.
species: dog
vuln: false
est: €680
Enrich
Point-in-time policy context, endorsements intact, partner data through EU-resident proxies. The next gate sees the full picture, with a replayable trail.
event: delay 6h
corridor: premium
territory: EU
Decide
Atomic decisions per gate, at the maturity level you have signed off. Every verdict signed, replayable and carrier-approved.
fault: display
cover: ext. warranty
age: 14 months
value: €1,299
region: DE
payout: €189
conf: 0.92
Measure the decision, not the claim.
A claim is not one decision. It is about 5 — coverage, liability, fraud, reserve, payout. Each one either clears on its own evidence or it doesn't.
Claim STP tells you a claim didn't finish. Atomic tells you which gate stopped it, on which corridor, today — so there is something to actually fix.
One low-confidence decision holds up a whole claim, so claim STP lags every improvement you make. Decision-level movement shows the same day.
Thresholds, authority caps and model versions are set per gate. You can tune a gate. You cannot tune an average of 5 of them.
Claim STP still falls out the other end — 62% today, 16pp behind the decisions that drive it. Worth reporting. Not worth steering by.
“Six weeks of manual enrichment, replaced with a signed decision trail our regulator could actually read. Claim STP went from 11% to 64% in a single quarter, and every edge case was traceable.”
Common questions.
Trust by architecture, not assertion.
Every decision signed, every model version locked, every data fetch logged. Compliance isn't bolted on — it's structural.
Also structural: Customer-managed keys · Canary and auto-rollback · Zero trust network · Pen tested quarterly · Human oversight gates above L2
Signed per atomic decision. Replayable against the exact model and prompt version that produced it. Medley AI keeps the trail, you keep the thresholds.
Pilot one corridor. Ninety days.
Fixed scope, fixed fee. A production reference at the end, not a proof of concept.