In September, we made a decision that most people would have thought was reckless.
We had a working product — DeRISK CRQ with its companion DeRISK QVM, deployed at over 300 industrial facilities across 2 continents. Tier 1 operators paying real money for OT cyber risk quantification. A market position most startups would protect at all costs.
But we decided to rebuild the company from the ground up.
New architecture. New team — from no agentic AI engineers to seven today. New market. Not because the old product was failing. Because the opportunity in front of us, and the mission to Build the Global Standard of Industrial Cyber Risk Quantification for Boards, Shareholders, Investors, Agencies, and the Risk Transfer Market was too large for the previous architecture to reach. Three things had converged: the market was finally understanding the cost of leaving critical infrastructure unprotected, agentic AI had matured enough to orchestrate complex specialist workflows, and we had one of the best OT cyber risk datasets in the market to ground it. This was not a reaction to a trend. It was a bet on a convergence we could see forming.
Since then, the team has been relentlessly and quietly building — the infrastructure to train agentic AI first, the product second. No announcements. No previews. Early adopters have tested it. Now it comes to light.
What we built is the DeRISK Intelligence Platform — a unified agentic AI architecture that serves industrials, insurers, and capital markets from one shared intelligence layer and one compounding data flywheel. Today, that platform ships its first product into the world, DeRISK UWA Agentic.
DeRISK UWA Agentic is an agentic AI underwriting platform built specifically for industrial cyber insurance — the segment of the market with the least actuarial data, the highest decision variance, and the most consequential risk.
It is not a chatbot. Not a co-pilot. Not a dashboard that summarizes a PDF.
It is an autonomous workflow. An Orchestration Agent coordinates five specialist AI agents through a defined underwriting pipeline — from submission to actuarial output — in ten to twenty minutes. The output is not a risk score. It is a complete, structured, actuarially grounded underwriting assessment: expected loss, maximum foreseeable loss, full loss exceedance curves, premium indication, a structured insurance program with recommended limits and retentions, and mandatory binding conditions with deadlines and consequences.
UWA Agentic is ready to ingest all standard questionnaires and applications used in the cyber insurance market today — including specialized OT supplementals and ransomware supplementals — across PDF, Excel, CSV, Word, and JSON formats. Whatever the broker sends, the platform reads it.
Every assessment is traceable. Every finding maps back to its source document, the pipeline step that produced it, and the model version that ran. A licensed underwriter reviews every output before any binding decision. The AI augments the underwriter — it never replaces them.
Three things no one else has built.
First, real-time OT threat intelligence. DeNexus maintains a proprietary cyber threat intelligence database, continuously updated — including MITRE ATT&CK for ICS techniques, which most platforms don't touch. On every single run, the insured's industry and geography are matched against active threat campaigns. The risk agents are pre-enriched with current threat context before they score a single control. If a new campaign surfaces mid-assessment, the pipeline interrupts and re-scores. This isn't static training data. It's live intelligence, injected at assessment time.
Second, dual IT and OT specialist agents. Two independent risk agents run in parallel — one scoring the IT control environment against industry-standard cybersecurity frameworks, the other scoring the OT environment against the leading international industrial cybersecurity standard. Neither agent sees the full picture alone. Cross-domain dependencies are detected: strong IT compliance can mask dangerous OT exposure.
And when OT-specific questions go unanswered — segmentation, safety interlocks, remote access to field devices — the system doesn't skip them. It doesn't average them. It flags every absence as a structural risk signal. We call this capability Sound of Silence. An underwriter scanning a hundred and thirty pages — the main application plus the OT supplemental — would likely miss the gap. UWA Agentic catches it. Every time.
Third, a proprietary actuarial engine — grounded in three hundred real industrial OT deployments since 2019. The only actuarial dataset built from native OT implementations at the account level. And it's modular: the insurer can substitute their own actuarial model entirely. All upstream findings feed in unchanged. Your pricing model. Our risk intelligence.
Two real submissions — processed during our early adopter program.
One with forty percent data complete — no industry sector, no revenue, no OT disclosure. Sound of Silence triggered. The platform produced a $266,000 expected annual loss, eight structured coverage lines, four mandatory binding conditions with deadlines, and a premium range of $79,000 to $114,000. 11 minutes and 47 seconds.
The other: a large European retailer with €40–50 billion in revenue and 262 million customer records. Three material underwriting gaps identified. OT security levels assessed. €9.2 million expected annual loss, seven coverage lines, three binding conditions, premium range of €3.1–4.0 million. 18 minutes and 42 seconds.
Incomplete data doesn't stop the assessment. It shapes it.
UWA Agentic is the first product on the DeRISK Intelligence Platform. There's more coming. I'll share details when the time is right.
What I can say is this: we didn't rebuild the company to launch one product. We rebuilt it to build a platform where every deployment, every assessment, and every threat cycle compounds the intelligence available to boards, shareholders, insurers, and the broader risk transfer market. UWA Agentic is the proof that the architecture works, that the intelligence is real, and that the market is ready.
BCG published their agentic AI maturity framework for insurance earlier this year. The framework defines five horizons: H0 (constrained, rule-based single task), H1 (single agent, multi-step but acts alone), H2 (deep agents, orchestrated specialists), H3 (role-based, end-to-end agent teams), and H4 (agent mesh, self-organizing — theoretical only). Most carriers sit at Horizon 0 to 1 — rule-based automation or single-agent tools. UWA Agentic deploys at Horizon 2 to 3 on day one — orchestrated specialist agents executing end-to-end workflows. BCG's explicit recommendation to the market: specialist insurers cannot build this in-house. Buy and integrate. We agree
Every component of UWA Agentic adapts to the insurer. The actuarial engine. The cybersecurity framework. The report structure. The questionnaire formats. The deployment model. The workflow integration. Pre-trained, not locked in.
Three tiers — triage, peer review, bind. Start small. Scale with confidence. No big-bang deployment. No headcount risk. The platform augments underwriters. It doesn't restructure underwriting teams.
If you underwrite industrial cyber risk — or if you're thinking about entering the market — I'd like to show you a live run. A real submission, processed in real time, with real output.
Before I close — thank you.
To our customers, who trusted DeNexus when OT cyber risk quantification was a category that barely existed — your confidence built the foundation everything else stands on. To the early adopters of UWA Agentic, who put a new product through real submissions and gave us the feedback that made it better. Chris Norwood, Cyber Analyst at Chaucer Group, put it simply: “DeRISK UWA delivers cyber risk insights for industrial sectors quickly and in a format underwriters can use to make informed decisions — or provide a deep dive into complex OT and property damage factors.” To my team, who rebuilt an entire company's architecture in under a year and never once asked whether it was possible. To my board, for trusting the vision when the safer path was to protect what we had. To my investors, for backing a decision that looked reckless on paper and turned out to be exactly right. And especially to Alex Horvitz, who helped me navigate the hardest transformation this company has been through. None of these ships without any of you.
The first product on the DeRISK Intelligence Platform is live. And there's more to come.