Customer & counterparty screening
One search box for onboarding, CSV/XLSX batches for portfolios — and continuous screening after that: every list update automatically delta-rescreens every affected customer. Screening history never expires.
Financial-crime screening & investigation platform
Eliquo screens your customers, payments, employees, and counterparties against sanctions, exclusion, and PEP lists — and returns decisions a regulator can read. Every answer carries its evidence, its calibrated probability, and a coordinate that replays it byte for byte, years later.
Built for banks, fintechs, and payment institutions with sanctions-compliance obligations.
Clears are explained the same way as matches — the closest rejected candidate, and why it fell short.
One platform, five jobs
One search box for onboarding, CSV/XLSX batches for portfolios — and continuous screening after that: every list update automatically delta-rescreens every affected customer. Screening history never expires.
A real-time evaluate API returns PASS or HOLD inside the payment flow — SWIFT MT, ISO 20022, Nacha IAT, and CSV. Stripped originators, missing cover data, and resubmitted rejects are caught by structure, not just by name.
Workforce and vendor rosters screen against SAM.gov, HHS OIG LEIE, and World Bank debarment through sector presets — compliance screening, deliberately bounded so it can never become a consumer report.
Documents in — deeds, registry extracts, scans included — an ownership graph out: directors, UBOs, vessels, identifiers, each claim quoting the page it came from. The OFAC 50% rule is computed over that graph.
Alerts group into cases with queues, SLA clocks, and maker-checker closure. A blocked payment starts a real 10-business-day OFAC report clock with an assembled packet. The examiner gets one exam-ready export.
eliquare (Latin, v.)
to strain clear; to refine a liquid until what remains is pure. The flow of customers and payments passes through, the designated few are caught, and everything else runs clear — with the evidence for why.
The matching engine
The system of record is an identity knowledge graph — full alias fans, shared identifiers, ownership edges, cross-list identity clusters — queried by embedding-first hybrid retrieval with a deterministic guarantee channel running in parallel.
Exact, character-equivalence, and certified-typology hits are MATCH regardless of what any model does. Learned layers can only add recall on top — the certification suite proves the floor holds with the models switched off. That is the structural answer to “what if the model is wrong?”
Every decision carries a per-feature evidence breakdown — alignment rules, token weights, embedding similarity, DOB and document corroboration — and a human-readable narrative. Thresholds are derived from your written error-rate policy, with the derivation as an audit artifact.
Ownership and control edges are parsed from the lists themselves, and a fixpoint computation designates entities majority-owned by sanctioned parties — entities on no published list, screened like listed names, with the full ownership path as evidence.
Every event embeds a run coordinate — list versions with release/download timestamps, index, model, rule artifacts, license and suppression sets. Replaying the coordinate reproduces the decision and its disposition byte for byte. Regulator replay is a built-in capability, not a data-recovery exercise.
Investigation & audit posture
True-positive, licensed-proceed, block, and reject closures require a second, different senior actor — self-approval is refused by conditional write, not by convention.
Recurring false positives are absorbed by adjudicated match-pair suppressions: four-eyes approval, mandatory expiry, pinned to the list version — and auto-voided the moment the list entry changes.
OFAC 10-business-day report obligations and SAR 30/60-day flags run on self-deleting scheduler timers with assembled report packets — never on a TTL that can lag 48 hours.
Every material event — ingestions, searches, verdicts, dispositions, approvals, deployments — lands in a hash-chained stream with a chain-verified export. One endpoint answers the examiner with one artifact.
REVIEW and NETWORK decisions are never auto-closed. Adverse media, PEP records, and free-text hits are capped at REVIEW by construction — they can cost analyst minutes, never a wrongful block.
Auto-closes are QA-sampled weekly and per-rule overturn rates land in a governance report — SR 11-7 outcomes analysis your validator can hold.
Editions & delivery
Serverless AWS end to end. Tenant data is isolated, users are your own, and cost scales with screening volume — zero idle cost when nothing screens.
Subscribe through your existing AWS bill: usage-metered screening, transactions, and connectors, with entitlements and lifecycle handled natively.
For banks whose raw data cannot leave their account: the data plane deploys in your AWS account under your KMS keys, Eliquo attaches cross-account, and deleting one queue policy is your kill switch.
Integration is a catalog, not a project: REST evaluate and S3 batch drop included; SQS, EventBridge, SFTP with PGP, Kafka/MSK, Kinesis, and an IBM MQ bridge blueprint — each priced up front, provisioned on demand, destroyed cleanly on deactivation.
Contact
Bring a test deck or a portfolio sample — we'll run it live and walk through every dossier, match and clear alike.
hello@eliquo.ai