Protos Labs is a Singapore-based cyber AI company founded by ex-Booz Allen cyber operators, focused on Adversary Risk Intelligence: analysing adversaries and the risk they pose to an enterprise. We have co-built and deployed Protos AI with defence and homeland-security agencies since 2024, where a wrong answer isn't an inconvenience, it's a mission failure. That's the standard we build to.
We started by putting a dollar figure on cyber risk, based on real adversary behaviour rather than generic frameworks.
But customers kept coming back to one part of it: the threat work. Threat feeds, vulnerability disclosures, dark web chatter, security alerts — all available, almost none of it tied to a specific organisation. What does this threat mean for us? Which of our assets, suppliers, software and data does it touch? What do we do first? Answering took analysts hours per incident — and it was exactly what executives and boards needed before they could decide anything.
AI moved that limit. The collection and analysis that once took a department could now run at machine scale, directed by the people who used to do it all by hand. So we rebuilt the product around one gap: the distance between a threat signal and what it means for your environment.
The work the product does — building a picture from incomplete data, working across sources, and producing conclusions that hold up under review — is the work our founders did for years before they started the company. We built the platform to do it at scale. We have active business in Singapore, the US and Thailand.
Protos AI was built on the hardest national-security problems, then brought to commercial enterprises. We spent over a year building, testing and refining the product in that setting before opening it to commercial customers in October 2025.
That shows in the product today. Sovereign deployment is built in, not bolted on later. Trust is measured, not asserted: the Trust Fabric scores every output on accuracy, reliability, consistency and speed, and every conclusion traces back to its sources and reasoning, because "the model said so" was never good enough for the teams we built with. The habit of working across multiple sources, finding needles in a haystack, came from analysts who work to that standard every day.
The result: things other vendors call roadmap — reliable outputs at scale, deployable in regulated environments, defensible under audit — were requirements for us from the start.
We don't name our partners. What we carry forward is how they work.
.png)