Early work on AI-assisted compliance systems
This early research explores how AI can support compliance teams by structuring evidence, surfacing uncertainty, and keeping regulated decisions under accountable human control.
From compliance evidence to controlled workflow decisions.
Compliance work depends on scattered information: policies, identity records, forms, documents, approvals, and correspondence. Anmel is studying how AI systems can organize that evidence, identify gaps, and support reviewers without turning regulated decisions into black boxes.
Turn policies, documents, customer records, and operational context into structured compliance evidence.
Route ambiguous findings, missing information, and control exceptions into human review workflows.
Preserve sources, model reasoning, review decisions, and approvals for reviewable compliance operations.
The goal is to make compliance operations more inspectable: what evidence was used, where the system was uncertain, which policy or control applied, and who made the final decision.
Built around evidence, controls, and accountable review.
Evidence layer
Policies, identity records, documents, correspondence, and transaction context organized into reviewable compliance cases.
Control workflow
Screening tasks, exception queues, escalation paths, reviewer assignments, approvals, and change history for regulated teams.
Oversight interface
AI-assisted summaries that surface relevant evidence, uncertainty, policy references, and the rationale behind recommended next steps.
This research became part of the foundation for Enmita-style systems: AI that helps regulated organizations understand documents, manage onboarding evidence, and keep compliance decisions traceable.
Return to updates