The NeuralMetrics Journey
The story of how NeuralMetrics rebuilt the front end of commercial underwriting — one milestone at a time.
“The gap between what a business says it does and what it actually does is the most expensive gap in insurance — and it can be closed.”
NeuralMetrics started with a single belief: the gap between what a business says it does and what it actually does is the most expensive gap in commercial insurance — and it can be closed. Every product since has been a consequence of that one idea.
Before deciding a risk, you have to truly know the business. So we built a business-entity intelligence layer drawing on 15+ real-world sources — licenses, permits, OSHA, EPA, NCCI, social signals — assembled county by county into 130M+ records, updated continuously. Classification is the commodity. This data is the moat.
We turned that data into precision the market couldn’t match: NAICS to 8 digits, plus SIC, ISO GL and WCC codes. Everyone agrees on the 6-digit code — the real exposure lives two digits deeper, and that’s where we underwrite.
Classification told carriers who the business was. ARKUS told them what could go wrong — surfacing and scoring the live exposures a questionnaire may not accurately capture, straight from public data.
Insight became a decision. GIA and GIA Lite score each risk against the carrier’s own appetite and underwriting guidelines, product by product — stopping out-of-appetite business before it ever reaches rating.
The last gap was the end of the policy period. CHRIS closed it — reconciling what was actually written against what should have been, and turning weeks of manual audit into automated validation.
This year the pieces became one system. Document Ingestion taught the platform to read. Book Roll scaled decisions from one submission to an entire portfolio. Hardened infrastructure and open APIs turned a precise engine into a full underwriting-and-audit platform.
One continuous pipeline — submission in, quote-ready decision out.
We’re no longer just a data company — we’re intelligence infrastructure for risk-bearing enterprises. Now NeuralMetrics speaks the Model Context Protocol, so any AI agent can call our intelligence directly. Bring your own agent; the intelligence is already there.
Provenance
One carrier asked. Many now use it — which is how we know the problem was industry-wide.
NeuralMetrics has never built a product to sell. Each one exists because a carrier brought us a problem they couldn’t solve with anything on the market — and we solved it with them: their guidelines, their data, their workflow. Then the next carrier turned out to have the same problem. And the next. Five products, five real problem statements — each one proof that what one underwriting team needed, the industry needed too.
Submissions were arriving without reliable class codes — and everything downstream depends on the code. We built classification that resolves the business off a live source: 6+2-digit NAICS, SIC, ISO GL and WCC, never a stale database. It became the layer every other product reads from.
Today, the entry point for every customer we serveThey could see the building, but not the businesses inside it — and in lessor’s risk, the tenants are the exposure. We built LRO to identify and classify every business at a shared address, turning exposure that was invisible at the property level into something visible at the risk level.
Now relied on by four national carriersThere was no fast way to read whether a submission was in appetite or out; the answer lived in PDFs and in people’s heads. We encoded their own guidelines and evaluated every submission against them, every source traceable — an eligibility decision that’s consistent between underwriters and defensible to an auditor.
Now used by four national carriersEven with the data in hand, submissions still needed an underwriter at every step. We built ARKUS to identify risk attributes by product and class and explain each one with live signals — surfacing exposures before the bind, not at the claim.
Now used by four national carriersPost-bind audit was still manual and email-driven — the last room in the building that hadn’t been automated. We built CHRIS to validate policy, payroll and secondary sources against the bound risk, closing the gap between what was written and what should have been.
In production todayBuilt for real use cases, then hardened through each new customer.
Start with one carrier’s genuine problem, solve it precisely, and let the market prove how universal it was. It’s also why a carrier can begin anywhere — a single data feed, a single agent — and expand across the rest as the same need surfaces in the next workflow.
Measurable outcomes from classification all the way through audit.
Exposures surfaced before bind, not after a claim.
The right code and the right appetite, so fewer risks get mis-rated.
Roughly 10 distinct risks surfaced per submission, on average.
Reconciling what was written against what should have been written.
From submission to audit — NeuralMetrics compounds accuracy and removes cost at every step.
We started by getting the data right. We’re finishing by putting that intelligence everywhere a risk decision gets made.
The guess is gone. What replaces it is us.
NeuralMetrics is a risk-intelligence platform for commercial insurance, powering classification, risk insight, eligibility, and premium audit for leading enterprise carriers.