Platform Governance Company Contact
Governance & deployment

An AI you can't audit is an AI you can't deploy.

CognitoClaims is engineered from the first line of code for the regulatory environment of Singapore and the region.

MAS FEAT PDPA MOH AI Guidelines Audit-ready by default
Governance by design

Compliance isn't a feature. It's the architecture.

MAS FEAT

Fairness, Ethics, Accountability, Transparency

Decision models designed and monitored against FEAT principles — with fairness testing across cohorts, human-in-the-loop controls on adverse outcomes, and clear accountability for every automated decision.

PDPA

Data protection, engineered in

Purpose limitation, consent handling, data-residency and retention controls native to the platform — personal health data handled the way Singapore law expects, with breach-readiness built into operations.

MOH AI Guidelines

Aligned with AI in healthcare

Built with reference to Singapore's guidelines for AI in healthcare: clinical safety, explainability, defined human oversight, and continuous monitoring of model behaviour in production.

Audit trail

Every decision, replayable

Complete, immutable decision trails on every claim — the clause applied, the benchmark referenced, the model version, the override. Ready for your internal auditors and your regulator, on demand.

Deployment

Your data never has to leave your perimeter.

Health-claims data is among the most sensitive data an organisation holds. Our deployment model starts from that fact.

Self-hosted domain models

Fine-tuned models deployed inside your VPC or on-premise. Full capability, zero data leaving your environment.

Integrates with what you run today

API-first, designed to sit alongside legacy core systems and TPA workflows — no rip-and-replace project.

Privacy-preserving by research

Active R&D in privacy-preserving computation for insurer–provider data exchange — collaboration without exposure.

Questions we get asked

Straight answers.

Does claim data leave our environment?

Not if you don't want it to. Our reference deployment runs the full engine — models, data, decisions — inside your VPC or on-premise. Nothing crosses your perimeter.

Can the AI deny a claim on its own?

No. The engine approves clean claims straight through, but adverse decisions are designed for human review — the AI prepares the analysis and the evidence; your assessor makes the call. Every override is captured in the same audit trail.

How do you handle documents in multiple languages?

Our ingestion models are built for the mixed-language reality of Southeast Asian healthcare documents. Fields extracted with low confidence are routed to human review — never silently guessed.

What does integration look like?

API-first. The engine sits alongside your existing core system, claims workflow, or TPA platform and takes work from it — you don't replace what you run today.

How do you prove a decision to an auditor or regulator?

Every decision carries an immutable, step-by-step trail: the policy clauses applied, the benchmarks referenced, the model version used, and any human involvement. It exports in an audit-ready format on demand.

Bring your compliance team to the briefing.

We built this for their questions. Seriously — bring them.

Request a briefing →