CEO Corner

A productive second day for Genellipse

We delivered more than 10 demonstrations and presentations of Cortex AI OS and AegisNow and arranged three additional follow-up meetings. I am pleased with the strong interest in Cortex AI OS and the opportunity to discuss where governed AI can make a meaningful difference to insurance organizations.

For me, the value of these conversations is the move from curiosity to a specific business problem: the work that takes too long, the information that is difficult to reconcile, or the decision that needs stronger controls. Our responsibility is to turn that interest into a partnership with clear priorities and measurable progress.

A question on the ITC agenda that deserves attention

This edition takes its title from ITC’s published September 30 session, presented in the programme by Partha Sen, Director — Insurance Solutions at Hexaware Technologies. Its focus on autonomous mobility raises a fundamental question: how should insurance respond when driving decisions increasingly depend on software? The reflections that follow are my own perspective on that question.

The opportunity extends beyond automating an existing insurance process. Insurers need to understand a changing source of risk, determine which evidence matters, and make coverage and claims processes understandable to customers and business partners.

When an accident happens, who is responsible?

The starting point is to establish what technology was actually operating. NHTSA distinguishes Level 2 driver assistance, where the driver must remain engaged and attentive, from systems that perform the driving task under defined conditions. Calling a vehicle “AI-enabled” does not tell us who was expected to do what.

An investigation may need to examine the actions of a human driver, the vehicle manufacturer, the automated-driving developer, the fleet operator, or a maintenance provider. Responsibility depends on the facts, the applicable jurisdiction, causation and the legal basis of the claim. It does not automatically transfer to a software company because an automated system was involved.

My practical question for insurers is whether they can reconstruct the event: which driving mode was active, whether the vehicle was operating within its intended conditions, whether a takeover request occurred, and which software version and maintenance history applied.

Well-resourced companies do not replace the need for evidence

The involvement of major technology and automotive businesses adds a commercial dimension to litigation. In my view, available insurance and the ability to recover compensation may influence whether an expensive, technically complex claim is worth pursuing. That is different from establishing legal responsibility: corporate wealth is not proof of liability.

Lawyers may examine product design, warnings, operation and maintenance, and investigate more than one party. Product-liability claims require a legally supported connection between the alleged defect and the injury; the precise requirements vary by jurisdiction and claim.

In US federal proceedings, discovery is subject to relevance, proportionality and applicable protections. Software history, testing records and incident logs may become important evidence. Reasonable preservation of relevant electronic information when litigation is anticipated is therefore an operational issue as well as a legal one.

I believe the best preparation starts before an incident: agree who retains critical records, who can access them, how their integrity is maintained, and how a preservation request reaches every relevant system.

Underwriting must consider shared technology exposure

A question I would put to an underwriting team is this: what happens when a single software release is used across thousands of vehicles? Even if individual journeys appear routine, a common dependency may create correlated exposure. Pricing and portfolio review should examine that possibility rather than assuming every vehicle represents an independent risk.

Insurers can ask about operating geography, weather restrictions, software-change controls, fleet maintenance, incident response and dependency on external services. Actuarial teams can explore scenarios and challenge assumptions, with uncertainty made explicit. These are considerations for assessment, not a claim that every autonomous fleet has the same risk profile.

The service experience matters too. An injured person should not have to resolve the entire technology supply chain before receiving clear guidance. Insurers need coordinated communication while coverage, responsibility and potential recovery from other parties are assessed.

The ecosystem is already taking shape

Waymo announced its Las Vegas public service in September 2026, while Zoox also offers a purpose-built robotaxi service in the city. They are useful examples of autonomous mobility becoming a practical insurance consideration. Their operating models and conditions should be evaluated individually.

On the insurance side, Marsh and Apollo’s ibott announced a facility for Uber’s autonomous ride-hailing partners in March 2026. My takeaway is that technology developers, operators, brokers and insurers need to design the operating and insurance arrangements together. These market examples are not Genellipse partnerships or endorsements.

Let us turn the conversation into a practical next step

The interest in Cortex AI OS at ITC is encouraging, and our three additional follow-ups create an opportunity to explore concrete needs with prospective partners. I look forward to continuing those conversations around AegisNow, governed AI and the data foundations that make both useful.

If these questions reflect your organization’s priorities, connect with me or the Genellipse team at genellipse.com, or email info@genellipse.com. Bring us a process that needs to work better, and let us establish together what a responsible, measurable improvement would look like.