Submissions arrive scattered
A single risk turns up as an application, a loss run, a SOV, a broker email and a handful of attachments — each stating overlapping facts in its own format, none of them authoritative.
Our mission
Commercial underwriting deserves an assurance layer
Commercial underwriting is still fragmented and largely manual. Risk arrives across scattered documents that quietly disagree with each other, and the contradictions and gaps that matter most are the ones most likely to go unnoticed until they are expensive. We build the layer that makes those things impossible to miss.
The problem
The information is there. The assurance is not.
Contradictions surface late
Two documents disagree on year built, construction, or square footage. Nothing in the workflow forces that disagreement into view, so it is found during a referral, an audit, or after a loss.
The file cannot defend itself
When a decision is questioned months later, the reasoning lives in inboxes and memory. There is no single record showing what was known, what was missing, and who decided.
The mission
Faster, more consistent, more defensible — decided by underwriters
We exist to help commercial insurers reach underwriting decisions more quickly, more consistently across a team, and with a record strong enough to stand up to scrutiny later. Every one of those gains has to arrive without taking the professional underwriter out of the decision.
That constraint is the design, not a disclaimer. The platform reads, reconciles, flags, and documents; it never concludes. Judgment about a risk belongs to a named person with authority and accountability, and our job is to make sure that person is working from a complete, verified, traceable file instead of a pile of attachments.
- Augment, never replace
- AI assists the file. The underwriter records the outcome, with a written rationale.
- Evidence over assertion
- Every fact traces to the document, excerpt, and page it came from.
- Defensible by default
- The trail is produced by doing the work, not assembled afterward for an audit.
The category
Underwriting assurance is its own category
Assurance is a distinct function: establishing that a decision was made on complete, consistent, verified information, by someone authorized to make it, with the reasoning preserved. Underwriting has never had a system dedicated to that. It has had systems for issuing policies, systems for managing relationships, and lately tools that read documents — and the assurance work has been left to discipline, spreadsheets, and memory.
That is the layer we build: between the submission and the decision, doing the reconciliation and record-keeping that no adjacent system owns. Being clear about the category also means being clear about its edges.
- Not a policy administration system
- We do not rate, issue, endorse, or service policies. The assurance layer sits before that system and hands it a decided, documented file.
- Not a broker CRM
- We do not manage relationships, pipelines, or production targets. Our unit of work is the submission and the evidence behind it.
- Not a claims platform
- Our scope ends at the underwriting decision. What happens after a loss is a different discipline with different records.
- Not an autonomous underwriting engine
- The platform never binds, declines, or refers on its own. Guideline signals are advisory and every outcome is recorded by a named underwriter.
- Not a document reader
- Extraction is a means, not the product. The value is what happens to the facts afterward: contradiction detection, gap detection, review, authority limits, and an immutable trail.
- Not a replacement for underwriting judgment
- The platform organises evidence and makes gaps and conflicts impossible to miss. Appetite, pricing, and the final call remain the work of a qualified underwriter operating within their authority.
A note on the company
We are an early-stage company, and we are building this because the problem is specific and unglamorous enough to be worth solving properly: the facts in a commercial submission disagree with each other, and nothing in the ordinary workflow makes anyone confront that before a decision is signed. We would rather build one layer that handles it rigorously than a broad platform that handles it incidentally. What is described on this site is what the platform does today.
Every commercial underwriting decision should be fast, consistent, and defensible — and still made by an underwriter.
See the assurance layer applied to a real submission: extraction with provenance, contradictions surfaced, gaps named, and a decision recorded under authority limits.