Monitor, investigate, document
Selected risk rules, source records, case decisions and an audit trail. Analysts propose; a different MLRO reviews.
Trade, payment and customer records often sit in separate systems. Marqib checks selected broker risk patterns, brings the supporting records into a case and helps your team record its decision and reporting draft.
Rules highlight selected patterns and keep their supporting records together. Your team checks the explanation and decides what to do.
Start with customer, trade and payment files. Optional adapters are tested with your firm’s accounts.
Rules and screening identify selected patterns. Each case shows the records and thresholds behind the alert.
Review the evidence and record the decision. An analyst’s proposal is reviewed by a different MLRO. Your authorised team checks and submits the reporting draft.
See what you can review in the pilot and what we are planning next.
Selected risk rules, source records, case decisions and an audit trail. Analysts propose; a different MLRO reviews.
Saved evidence versions, data-quality controls and changed-proposal review have local checks. Release and final verification are pending.
Trading, KYC and crypto adapters need validation with your own accounts. Reporting exports are drafts.
Feed reconciliation, document requests, deeper relationship review and rule observation before decisions.
Parameters are configurable and can be tested on loaded history. Data availability, thresholds and reporting fields are agreed and checked during the pilot.
Connectors, webhooks and files for the systems brokers and payment firms already run.
Beta connectors are validated with your account during the pilot. All integrations and their status →
Regulated firms are right to ask what leaves their environment and what a model is allowed to decide. The answers are designed into the platform rather than left to policy.
Whether a matter is flagged is determined by rules and statistics that the firm can inspect and tune. A language model never decides what is suspicious.
Where enabled, a model drafts from the rule’s evidence table. Valid evidence references are checked, but that does not prove that a sentence is true. Rule text and model text are shown side by side for human review.
Supported identifiers are replaced by tokens before an optional model request. This is pseudonymisation, not a guarantee that every free-text detail is anonymous. The firm reviews the payload, provider and permitted data flow before enabling drafting.
The same platform operates with a model endpoint in the firm's region (for example Azure OpenAI in the UAE or AWS Bedrock in Bahrain), with an open-weights model on the firm's own servers, or with no model at all, in which case the rule-generated draft is the narrative.
Provider, model, prompt hash and the number of pseudonymised tokens are written to the audit log for each draft.
app.marqib.com runs on synthetic data in a vendor-hosted environment so that prospects can see the product. A pilot is never run there: it gets its own deployment and database, in the firm's own cloud account or on a dedicated instance we operate for that firm alone, with the model endpoint chosen by the firm.
The detection engine is common to all deployments; the filing adapter and typology thresholds follow the jurisdiction. The goAML export is a schema-shaped draft: indicator codes and party fields are mapped to the FIU's current list during the pilot, before any live submission.
Define a broker use case and representative records, then agree installation, rule checks and reviewer responsibilities. Measures such as review time and reporting turnaround are agreed targets to evaluate, not measured results on this site.
Scope and fee agreed before starting
Timing and commercial terms are set in the written pilot proposal.
Marqib is built by former senior managers of licensed brokerage and investment firms who have run regulated businesses across several jurisdictions, sat on the boards of regulated subsidiaries and answered to regulators for the anti-money-laundering controls this platform automates. The product is shaped by the review work an MLRO actually does, not by a generic model of it.
Marqib is developed by Nordera Works.