Who should participate
A bank or payment firm evaluating scam review and local transaction monitoring. Bring a compliance owner, an analyst or operations lead, and an IT contact. Agree the evaluation question together.
The ZQUAS pilot
Updated: 7 October 2026
Start inside
your own institution.
A 12-week local pilot for scam review and local transaction monitoring on your own data and infrastructure. An optional network phase can follow after you review the local results.
TRL 6. Synthetic data only. No production deployments.
Before you begin
Five practical answers.
What your team supplies
An agreed local dataset and its permitted use, account and payment history, and the outcomes available for evaluation. Your team provides access to suitable infrastructure, including an NVIDIA GPU, and completes its own security, legal and procurement checks.
What ZQUAS supplies
The pilot software, support to configure scam review and local monitoring, the signed decision log and its verification check. We agree the setup, evaluation support and report with your team before the pilot starts.
What gets measured
Scam payments caught, at most 1 review per 500 payments and 1 hold per 1,000, analyst workload, local monitoring workload and coverage, verification of signed decision records. These are evaluation measures and limits, not promised detection rates. Local transaction monitoring selects investigation work and does not itself hold payments.
What happens afterwards
Review the evidence, limitations and operational fit together. Decide whether to continue locally, change the scope, or stop. An optional network phase requires a separate agreement on participants, sharing policy, security, legal basis and success measures.
The local phase
12 weeks, with a clear review at the end.
The local pilot starts after scope, agreements, access and data readiness are confirmed. The plan below is agreed with your team before the start.
- Weeks 1 to 2
Agree the scope
Confirm the evaluation question, local data, access, responsibilities and success measures.
- Weeks 3 to 4
Configure the local pilot
Prepare local inputs and configuration for scam review and transaction monitoring. Check the decision log and verification process.
- Weeks 5 to 8
Run the local evaluation
Evaluate scam review and local transaction monitoring. Review nominated work with analysts. Record outcomes, monitoring coverage and workload against the agreed limits.
- Weeks 9 to 10
Assess the results
Examine scam payments caught, adherence to review and hold limits, analyst workload, local monitoring workload and coverage, and signed-log verification.
- Weeks 11 to 12
Make the next decision
Document results and limitations. Decide whether to continue locally or scope an optional network phase.
Data sovereignty
Your records stay with you.
The prototype supports bounded information sharing and signed decision records under the stated trust model.
Network participation is optional and follows the local phase by separate agreement. Current federation security is semi-honest: it assumes participants follow the protocol.
Evidence today
A prototype with results to examine.
The prototype supports bounded information sharing and signed decision records under the stated trust model.
ZQUAS governs decisions made by automated systems and AI agents. Financial crime is where it starts.
Programmes and sandboxes
Testing and engagement.
ZQUAS is taking part in the FCA's Supercharged Sandbox. The FCA does not endorse ZQUAS, its products or its results.
The next step
Explore a local pilot.
12 weeks inside your institution. An optional network phase follows only by agreement.