Nexusdesk.io

Nexusdesk Pte Ltd · Singapore

Make AI provably safe for everyone.

Architecture, not promises.

Regulated industries cannot adopt AI fast enough. They also cannot afford to adopt it wrong. Nexusdesk builds the architectural proof that lets them do both at once.

The problem

Regulated firms are adopting AI faster than they can prove what it did.

Financial services, healthcare, legal. Every regulated industry is racing to put AI inside core operations. Custom assistants. Internal copilots. Agents touching client files, case histories, financial records.

No regulated firm today can produce a verifiable record of what crossed the line between its systems and the AI providers its staff use. When a regulator, an insurer, or a client asks, and they will, most firms will not have an answer.

The window is short. Regulators are publishing AI frameworks. Insurers are pricing AI risk. Boards are asking the question. The firms that solve their AI accountability problem in the next 24 months will move faster than the ones that do not. The ones that do not will end up in someone's enforcement statistics.

Our thesis

Compliance is moving from a promise to a property.

For decades, regulated industries managed technology risk with policies, attestations, and audits. That model breaks when the technology is AI: non-deterministic, fast, and built outside the firm. Nexusdesk is built on a thesis: the only durable answer is architectural. Proof built into the system, not promised by the operator. Our first product, Evidor, is the proof of that thesis: a compliance plane between regulated firms and the AI providers their staff use. It classifies and redacts what cannot leave the firm, and produces a tamper-evident record of every prompt sent and every response returned.

See Evidor

The team

Built by operators from compliance and deep-tech.

Anson Zeall

Founder

Compliance operator and computer scientist. Spent his career getting fintech and crypto firms through the world's hardest regulators. Co-founder and chairman of ACCESS, now building the proof layer for regulated AI.

Chia Soo Hian

Technical advisor

Masters in Engineering, NUS. Two decades building enterprise systems inside regulated environments.

What we have shipped and filed

Prototype, patent, entity.

Working prototype shipped. Classification, redaction, hash-chained audit. See how it works

US provisional patent filed. 63/885,250.

Singapore-incorporated. UEN 202528726G.

Get in touch

Three reasons people email us. Three places to start.