Financial services
Transaction review, AML screening, and escalation chains all share the same requirement: not just that a decision was correct, but that it can be reconstructed afterwards, independently, without relying on the institution's own logs. This page sets out how such a workflow would be planned and run on inTheEU Assist, not a record of a completed deployment.
How the workflow would be structured
A proposed multi-step action, a transaction review, a flagged pattern, an escalation, would be planned as a chain of stages. The full sequence is hashed before any of it runs, so the plan an approver reviews is exactly the plan that executes. There is no gap between what was shown to a human and what the system actually did.
Stages flagged as risk-bearing (an escalation, a customer-facing output, a final decision) would carry an independent, second approval, separate from the initial plan-level sign-off. A general approval for the workflow does not substitute for a specific approval at the point that actually matters.
Where human approval sits
Every stage that could affect a customer, a transaction, or a regulatory filing would require an explicit signed approval before proceeding. Compliance and quant teams remain the authority on what constitutes a risk signal. The platform's role is to make sure that once they decide, the decision and its basis are provable, not just recorded.
What this draws on today, and what it does not
The two-tier approval structure, the plan-level hashing, and the signed, offline-verifiable audit trail behind every execution step are working parts of the platform. What is not part of this: any pre-built AML or fraud-detection model. The institution's own compliance and quant teams would supply the logic that defines a risk signal; the platform supplies the governed execution and audit layer underneath it.
Some of the architecture underlying this capability set is documented in filings still in progress. This page describes what the platform is designed to do, at a level appropriate for planning a pilot, not an implementation specification.
Platform features and interface are subject to continuous improvement. Descriptions in this guide may not reflect the exact layout or wording of the latest released version.