Kalepa

The removal test leaves an empty screen. Copilot is a workbench whose contents are produced by models: the structured read of a messy submission document, the cross reference against billions of external data points, the exposures the broker did not disclose, the priority score against appetite. The company states the product works on day one with no integrations, which means there is no system of record beneath it supplying anything.

This is the cleanest contrast in this pocket with the orchestration shaped vendors assessed alongside it, where a working workbench predated the models and would survive their removal. Kalepa had no product before the models.

Three named mechanisms, each with a stated position in the path. An ensemble engine selects the model or method for a task in real time and is described as knowing when to automate rather than amplify underwriter judgement, so the automation level is itself a decided property rather than a setting. Independent judging agents validate critical extracted fields, continuously measuring confidence and catching errors before output reaches the underwriter.

And the division of labour is stated plainly and narrowly: Copilot reads, cross references and scores, the underwriter decides, and the reasoning behind every flag is visible. Recorded as a limitation rather than a deduction, because it applies to the whole verification approach and not to this vendor alone: the judging agents are independent of the extraction agent but not of the model class, so a failure mode shared across both passes verification unnoticed. That is the same pattern this index flagged in health artificial intelligence self verification, and it is worth asking of every vendor that validates model output with another model.

The architecture is the evidence and it is genuine product design rather than assertion: independent judging agents validate critical fields, confidence is measured continuously rather than sampled, and the company frames consistency at scale as the objective. That places it alongside the product design B grades this index has given for glass box attribution and per tenant model isolation. What holds it there is a gap the vendor draws attention to itself.

Kalepa claims market leading underwriting accuracy and verifiable results, and publishes no accuracy figure, no error rate, no benchmark and no independent evaluation. An unquantified superlative about accuracy is the same shape as a compliance vendor advertising a large cut in false positives while publishing nothing about the cases it misses. The measurement infrastructure described here is exactly what would produce the number, which makes its absence more conspicuous rather than less.

Three named carriers spanning distinct shapes: Munich Re Specialty, Bowhead and SECURA, which covers reinsurance specialty, excess and surplus, and regional mutual business. A published case study on a mid market division of a top 15 global property and casualty carrier reports that within two months time to quote fell 40 percent, bind rate rose from 30 percent to 34.5 percent, and compliance with appetite and guidelines passed 95 percent, achieved with 15 percent fewer underwriters on staff.

A separate figure claims complex risks quoted 58 percent faster. Recognised in the InsurTech100 four years running. The gap that stops this being the strongest evidence in the pocket: the carrier carrying the hard numbers is described rather than named, and the three named carriers carry no figures, so the join between the two is missing.

The company states that Copilot automatically learns what the best underwriters are doing, which makes the cross customer question unavoidable rather than theoretical, and it is unanswered. Whose underwriters, and whether the behaviour learned inside one carrier's deployment informs scoring or prioritisation served to another, is not addressed anywhere located.

The platform also aggregates third party and open web sources into a single view, and nothing describes how source reliability is assessed or what happens when an external record is wrong. United States server location is stated and is credited on the deployment axis, but a data location is not a stewardship position.

A privacy policy is published and no data processing agreement, subprocessor list, retention schedule or lawful basis statement was located. Exposure would normally be low for a commercial lines product, since the subject of a submission is a business. It is higher than that here because the platform deliberately retrieves criminal activity and pending legal action records, which concern identifiable individuals connected to the insured business, from third party and open web sources. Material of that kind attracts consumer reporting and data protection questions in several jurisdictions and nothing public addresses them.

A service organisation control type two report is held and stated clearly on the product page rather than buried in a footer, alongside a commitment that data sits on United States based cloud servers. Type two is the audited operating effectiveness report rather than the point in time design report, which is the version enterprise buyers want.

Held at the lower end of this grade because it is one certification standing alone: no ISO 27001, no artificial intelligence management system certification, no trust centre, no subprocessor list and no penetration testing statement. That is materially behind the peers assessed alongside it in this pocket, two of which hold three audited certifications each, and a carrier running a vendor security review would notice the difference.

Kalepa is a technology supplier to carriers and managing general agents and holds no insurance licence, which is the correct posture and carries no penalty. No statement of regulatory position was located in any form: no classification under the European artificial intelligence regime, no engagement with insurance supervisors, no sandbox participation, no programme admission and no supervised test. Compliance with appetite and underwriting guidelines is a product capability measured against the carrier's own rules and should not be read as regulatory standing.

Nothing on fairness testing, differential outcomes or governance review was located, and the exposure here is more concrete than for most commercial lines vendors in this index. Copilot surfaces criminal activity and pending legal actions from external and open web sources as underwriting exposures, and the vendor's own case study describes the target exposures as violent incidents at insured premises.

Screening businesses on crime data attached to a location, in a United States market, imports whatever geographic patterns that data carries, and two otherwise identical restaurants in different postcodes will not screen alike. Held at C rather than lower because the insured is a business rather than a consumer, which weakens the protected class analysis considerably, and because state unfair discrimination duties fall on the carrier rather than the supplier. The silence is still the sharpest bias gap found in this pocket.

No error rate, remediation commitment, liability position or correction path is published. The recourse gap here has a specific and uncommon shape. Copilot pulls criminal records, legal actions and open web material about a business into the underwriting file, and a business declined or surcharged on the strength of a third party record it never sees has no route to inspect or correct it.

The confidence and judging agent architecture guards against Kalepa misreading a document, which is a control against the vendor's own error. It does nothing about an upstream source being wrong, and that is the failure mode most likely to reach an applicant.

The ensemble engine is described as selecting the right model or method in real time, which states plainly that multiple models are in use and names none of them. No provider, model family, version or licensing arrangement is disclosed, and the claim that the models are best in class carries no supporting detail.

Partial credit is due for one element other vendors omit entirely: the United States server commitment establishes country of processing for the data even though it says nothing about which models process it.

Integration depth is deliberately traded away for speed, and the company says so: Copilot works out of the box on day one with no integrations, with email inbox, portal and policy system connections available afterwards. That is a real commercial advantage against implementations that take a year, and it is the opposite of what this axis measures. No policy administration system, rating engine or workbench is named as a supported integration, and no integration partner is named at all.

Compare the peers assessed alongside it, which name specific core platforms, specific rating engines and specific data providers. A carrier adopting Copilot gets value quickly and does not get it wired into the systems where the policy is eventually issued.

Kalepa answers the question almost every peer in this pocket leaves open: all data is stored on United States based cloud servers. That is an actual residency statement rather than a gesture at one, and it is the only one found across five insurance platforms assessed in this session, including several far larger. Held off the top grade because the answer is single valued and the company operates across more than ten countries.

No European or United Kingdom region, single tenant option or self hosted path is offered or mentioned, so for a carrier subject to European data requirements the clarity of the disclosure is what reveals the problem rather than solving it.

No pricing, tier structure, billing basis or indicative range is published, and the route to a number is a demo request. This is the index norm and is measured against Sumsub, which publishes per verification rates on a public page.

The published case study gives a carrier every input for a return calculation except the cost: a 40 percent reduction in time to quote, a bind rate moving from 30 to 34.5 percent and 15 percent fewer underwriters are all quantified, with no price to set against them.

Commercial and specialty property and casualty carriers and managing general agents, with the named client base spanning reinsurance specialty, excess and surplus and regional mutual business, which is a wider spread of carrier type than the company's size would suggest. Line coverage handles multi location, fleet and additional insured complexity rather than simple risks only.

Held off the top grade on the standard applied across this pocket: life, health and personal lines are absent, so this is depth within one half of the insurance market. Scale is also a factor, with roughly 54 staff and a Series A balance sheet against peers serving hundreds of billions in premium.