One tier ladder. Six stages. Two product lines flow through it.
Retroactive Verification (looking backward at historical payments) and Pre-Settlement Attestation (looking forward at pending payments) are different products. They sell to different buyers with different urgency. But they run through the same six-stage tier ladder - and that's the point. The ladder is the operating system; the product lines are what gets routed through it.
Read top-to-bottom. Both products. One ladder.
Two product lines enter at the top. Each one runs through the same six stages. Tiers are paid steps the customer authorizes. Ava agents are autonomous reasoning layers between tiers - they decide what to escalate, what to bundle, and what stops here. By the bottom, the case is either closed (no fraud found, sealed) or carried into court-ready evidence with continuous escalation.
Tiers are billed. Customer authorizes each one explicitly. Ava agents are included between tiers - they're the agentic reasoning that decides what's worth paying for next, not the paid event itself.
What happens at each step.
Tier 1 · Flat-fee detection scan
Customer ships a batch of cases (claims, transits, payments). JIL runs the 175-check catalog against free public data: OFAC, UN, EU, UK sanctions; CMS PECOS + ownership + LEIE; SEC EDGAR; FINRA BrokerCheck; CFTC enforcement; NPPES NPI registry; SAM.gov exclusions; OpenSanctions; GLEIF; RDAP; Etherscan label cloud; ScamSniffer; OFAC SDN crypto; DefiLlama; FATF jurisdictions; and more. Output is a deterministic findings list: which checks fired, with what severity, against which source, with provenance. Pricing is flat per scan (volume-tiered).
Ava · Agentic AI gating
Ava reads the Tier 1 findings and decides what's worth paying to substantiate. If 47 findings fired and 35 of them are textbook clean-with-paper-trail (already-known exclusions, no contested elements), Ava closes them at Tier 1 with a sealed record. The remaining 12 it groups by counterparty / scheme / shared infrastructure and recommends a Tier 2 plan: which paid sources to query, in what order, expected cost, expected substantiation rate. Customer approves or trims the plan before any commercial-sub charges hit.
Tier 2 · Paid substantiation
JIL pulls in the commercial subscriptions and customer-supplied feeds the Ava plan called for: Bloomberg market reference, Refinitiv (LSEG), Plaid banking, IRS IVES, ATTOM property, Etherscan Pro, Helius RPC, Chainalysis KYT (where customer rides their own), TRM Labs, OpenCorporates, plus any customer-internal datasets. Human analyst review on flagged complex cases. Output is a substantiated case file: each finding from Tier 1 is now either confirmed-with-evidence or cleared-with-evidence; gap report identifies what couldn't be substantiated and why.
Ava Pro · Litigation-prep agent
Advanced agentic reasoning over the substantiated case file. Runs on AWS Bedrock + SageMaker. Builds the case-of-fact: chronology, counterparty graph, dollar exposure, statutory framing (False Claims Act, BSA, OFAC, MTL, state-specific). Identifies remediation paths (settlement, voluntary disclosure, qui tam). Drafts the CREB™ outline. Decides whether the case clears the Tier 3 bar (court-ready) or closes at Tier 2 with a sealed substantiated record.
Tier 3 · Court-Ready Evidence Bundle (CREB™)
The filing-grade deliverable. CREB™ is sealed, signed by the JIL validator quorum, anchored to CourtChain™ (FRE 902(14) admissible without live testimony), reproducible from canonical bytes. Contains the complete evidence chain: every Tier 1 finding, every Tier 2 substantiation step, every Ava reasoning artifact, every source attestation, the case-of-fact narrative, the recommended remediation path, all signing-officer attestations. Per-case priced; designed to be the one exhibit a counsel hands a regulator or a court.
Ava Pro+ · Continuous escalation
After CREB™ delivery, Ava Pro+ continues. Routes the CREB™ to client counsel, the relevant regulator, and (where the customer authorizes) law-enforcement. Tracks downstream events: settlement filed, indictment unsealed, recovery wired, statute-of-limitations countdown. Updates the case file as new public information surfaces. Subscription-priced, separate from the per-case Tier fees.
Retroactive looks back. Pre-Settlement looks forward.
Retroactive Verification
The customer ships historical paid claims (Medicaid, Medicare, commercial, corporate). The ladder runs backward from settlement: Tier 1 detects irregularities in the existing payment history; Ava decides which warrant paid investigation; Tier 2 substantiates with commercial sources + analyst review; Ava Pro builds the case-of-fact; Tier 3 produces a CREB™; Ava Pro+ routes to counsel + regulator + tracks recovery.
Buyer urgency: the CMS 60-Day Rule. Once an MCO has identified or had reason to identify an overpayment, they have 60 days under the False Claims Act to report and return - or they're knowingly concealing. This ladder gives them documented detection + substantiation + recovery evidence inside the window.
Pre-Settlement Attestation
The customer ships pending payments / pending claims / pending crypto transits BEFORE they clear. The ladder runs forward of settlement: Tier 1 screens at submission time; Ava decides whether to clear, hold for paid investigation, or auto-fail; Tier 2 substantiates the holds; Ava Pro builds a case-of-fact for any that need formal action; Tier 3 produces a CREB™ on the worst; Ava Pro+ keeps watching downstream.
Buyer urgency: Loss prevention. Pre-Settlement interdicts the bad payment BEFORE it leaves. For an MCO routing claim adjudication, a custodian routing a transit, or a federal agency routing a grant disbursement, this is the moment to catch the issue - not after the wire is irreversible.
Same ladder. Same engine. Different time horizon.
The architectural insight: the engine doesn't care whether the case is historical or pending. It cares whether the case meets the criteria at each tier. Same 500+ checks. Same Ava agents. Same Tier 2 commercial-sub stack. Same CREB™ format. Same Ava Pro+ tracking. The customer chooses which door to enter; the ladder is the same.
All eight run on the same ladder.
Each capability is a class of detection check. Each is independently substantiable from public-data evidence.
1 · Claim-pattern anomaly
Long lengths-of-stay, atypical billing volumes, statistical outliers vs cohort baseline, shared codes across providers.
2 · Multi-state UBO ring
Same beneficial owner appearing across 13 states under different LLCs, same tax IDs cycling through different entity names.
3 · Bank-fingerprint clustering
14 LLCs sharing 2 bank deposit fingerprints. Real organizational structure hidden behind nominally-distinct entities.
4 · Strip-mall DME
Durable medical equipment provider operating from a strip-mall storefront billing for 47 wheelchairs/month. Premise mismatch + volume.
5 · Premise-business mismatch
Therapy practice billing from a daycare premise; medical billing from a residential address; lab work from a mailbox-store address.
6 · Phantom telehealth
Provider billing 18 hours/day of telehealth sessions per single rendering provider. Time-budget impossibility.
7 · Excluded-provider reactivation
OIG-excluded provider resurfacing under a new tax ID, new entity name, but same physical address + same beneficial owner.
8 · All-anchors-firing case
The flagship: all five anchors fire at once - UBO ring, bank fingerprint, premise mismatch, billing pattern, OIG history.