The challenge
StealthIQ is an intelligent risk analytics platform for financial institutions, covering real time transaction monitoring across ACH, wire, card and crypto channels, machine learning based fraud detection with network graph analysis, synthetic identity and account takeover prevention, compliance automation across BSA and AML, KYC, SAR filing, sanctions screening, and explainable credit and portfolio risk scoring.
The scoring engine returns risk verdicts in under fifty milliseconds, which is precisely why a language model does not belong in that path. The opportunity sat on either side of it: the analyst work that consumes a score, and the document work that feeds one. Alert triage and investigation, KYC and onboarding document review, regulatory narrative drafting, and analyst facing explanation of model output were all candidates.
The constraint was not model access, it was team capability. Consistent prompt design, outputs conforming to the schemas the case manager and SAR generator already consume, an evaluation method that catches regressions before release, and defensible handling of customer data in prompts and logs. In a supervised environment, an unreviewed model output in a filing path is a compliance event rather than a bug.
What Kriv AI did
Discovery first, then a calibrated curriculum
Week one established what would actually be taught. Working sessions covered current and intended Claude usage, target workflows, output schemas, analyst review points and known failure modes, alongside a review of customer supplied non sensitive prompt samples, schemas and typologies. Three priority workflows were confirmed in writing as the teaching cases, recorded with the participant roster, schedule and success measures in a calibration memo the customer accepted.
Twelve hours, six sessions, exercises inside the session
Instruction ran remotely to a cohort of five to ten named participants from engineering, product and operations. Hands on exercises were worked inside session time rather than scheduled as separate labs, which is what keeps attendance and completion aligned in a team with live operational duties.
- Claude fundamentals for risk and compliance operations, including the boundary between deterministic rules, the existing ensemble scoring, and language model judgment.
- Prompt engineering for investigative accuracy on synthetic and de-identified case text: typology handling, ambiguity and refusal behaviour, and reducing unsupported or over confident output in analyst facing text.
- Structured outputs and tool use: schema conformant JSON for the downstream case manager, alert engine and filing path, tool definitions, multi step orchestration, and retrieval over policy and typology libraries.
- Document and entity analysis at scale: long document analysis for KYC and onboarding packets, entity and counterparty extraction, adverse media and sanctions hit disposition support, and citation and traceability for every extracted fact.
- Evaluation and quality control: test sets from de-identified cases, scoring rubrics for factuality and completeness, regression testing, human in the loop review design, and false positive against false negative trade offs.
- Safe deployment in a regulated environment: minimization of customer data in prompts and logs, prompt injection and untrusted input handling including adversarial content in transaction memos and uploaded documents, audit trail and explainability expectations, and cost and latency management.
No customer data, written into the contract
Kriv AI did not request, receive, access, store or process customer personally identifiable information, nonpublic personal information, account or transaction records identifying real individuals, or cardholder data, and held no access to production systems, data stores or credentials at any point. Every teaching example ran on synthetic or de-identified data the customer supplied or approved, and all exercise work was executed by StealthIQ participants inside StealthIQ's own environment. The engagement also explicitly excluded model risk management documentation, any determination of the regulatory sufficiency of a filing or alert disposition, and any modification, benchmarking or validation of StealthIQ's existing models. Those remained StealthIQ's responsibility and the contract said so.
Results
- All twelve instruction hours delivered on schedule across three consecutive weeks, with no change orders.
- A cohort of engineering, product and operations staff able to design, test and iterate Claude prompts for risk, fraud and compliance workflows without external assistance.
- All three priority workflows documented with working prompts and stated limitations in a reference pack, accepted against written criteria.
- A handover pack the team keeps: the prompt and pattern reference in repo ready Markdown, editable training materials for all six sessions, session recordings, and a closeout summary.
- StealthIQ uses Claude in day to day operations against the workflows established during the program, operating independently at handover.
About StealthIQ
StealthIQ is an intelligent risk analytics platform for financial institutions, covering real time transaction monitoring, machine learning based fraud detection, compliance automation across BSA and AML, KYC and sanctions screening, and explainable credit and portfolio risk scoring. For more information, visit stealth-iq.com.