Let’s Talk AI Operating Systems With Earnix

Insurance Edge caught up with Robin Gilthorpe, CEO at Earnix, to learn more about the AI OS concept the company is launching. Having AI sit as an actual operating system, rather than utilising it as an add-on in existing systems or data lakes, might just make more sense for many insurance brands.

IE: AI has really gone from development into deployment for insurers in the last year or two, hasn’t it?

RG: Yes, and that shift is exactly what we are seeing with insurers right now. The market is under pressure from rising risk, regulatory scrutiny, changing rate environments, margin pressure, and higher customer expectations. That is creating real demand for AI that can move beyond experimentation and into production.

The next phase is not about isolated AI pilots or dashboards. It is about operationalizing AI inside the workflows and decisions that drive performance across the insurance business. Insurers need to make faster, more precise, and more governed decisions across pricing, underwriting, claims, customer engagement, and retention.

That is why AIOS is so important. It is the AI Orchestration System for Insurance, built to connect the data, models, agents, workflows, governance, and human oversight insurers need in one decisioning environment. Insurance brands need intelligence they can trust, especially when decisions involve risk selection, rates, regulatory requirements, and customer outcomes.

The industry has spent years digitizing processes and modernizing systems. The next chapter is about continuous adaptation: using AI to respond to changing risk, market conditions, and customer behavior in real time, without forcing insurers to rip and replace the systems they already rely on.

IE: One topic the industry is talking about right now is orchestration. Having an AI conductor for all the tech and systems that have been developed. Do you see that being a trend now?

RG: Yes, and the conductor analogy works well. Insurers already have many of the instruments: models, data sources, policy systems, underwriting platforms, claims systems, customer systems, and now AI agents. The challenge is making all of those capabilities work together in a coordinated, governed way.

That is what orchestration is really about. It is not simply adding AI to a dashboard. It is bringing intelligence into the decisions themselves, so the business can act faster and more consistently.

There are three pillars that matter. The first is speed: moving AI from analysis into everyday workflows, so decisions that once took days or weeks can happen much faster. The second is orchestration: connecting models, data, agents, workflows, and governance across functions that have historically operated in silos. The third is trust: making sure AI is explainable, auditable, monitored, and appropriate for a regulated environment.

That last point is critical. In insurance, AI cannot be a black box. Some decisions can be automated, some should be assisted, and some will always require human judgment. AIOS is designed to support that range, with governance and human oversight built in.

IE: In the future, some scenarios will be flagged up, like potential fraud in claims, but with the AI system suggesting that human intervention might be needed. Can it help compliance in every major market and help build consumer trust as well?

RG: Yes, and that is why insurers need a dial, not an on-off switch. Different insurance lines, products, markets, and customer moments require different levels of automation and human oversight.

A straightforward transaction may be appropriate for a high degree of automation. A complex specialty risk, a sensitive claim, or a regulatory edge case may require human review. The system needs to support both, with clear governance around when AI can act, when it should recommend, and when a person needs to step in.

That is also why this cannot be a rip-and-replace moment. Insurers have made major investments in core systems, data infrastructure, underwriting platforms, claims systems, and customer tools. AIOS is designed to work above and across those existing environments, connecting them into a governed decisioning layer rather than forcing insurers to start over.

For compliance and consumer trust, explainability is essential. Insurers need to know what data was used, what model or logic informed the decision, what action was taken, and where human oversight was applied. That level of traceability is what makes AI usable in a regulated industry.

IE: The insurance sector has learned from the legacy vs existing systems era, AI deployment can’t be another battle between the old and new, can it?

RG: No, it can’t. One of the lessons from the last decade is that insurers cannot afford another rip-and-replace cycle. They have made major investments in core systems, data environments, underwriting platforms, claims systems, and customer engagement tools. AI has to enhance those investments, not force insurers to start over.

That is one of the reasons orchestration matters. AI in insurance has to operate across the existing technology environment, with the right governance, explainability, and human oversight built in. For executives, the concern has never been whether AI is powerful. It is whether AI can be trusted in regulated, high-stakes decisions.

The orchestration approach helps address that. It gives insurers a way to connect models, data, agents, workflows, and business rules while maintaining control over how AI is used. That is where the “dial” becomes important. Some processes can be more automated. Others need human review. The system has to support both.

We have seen a couple of years of AI pilots across the industry. Pilots have been useful, but the real question now is how insurers move AI into production and capture measurable value from it. That is the exciting part: delivery at scale, inside the workflows and decisions that actually run the business.

IE: In some respects, is China leading the way precisely because they are using AI at scale in real-world projects, like the building of new railway stations? Robots monitor site safety, AI helps engineers calculate every detail of the massive supply chain, weather windows, materials needed and so on. Are there lessons for insurers in claims after Cat events when we see China using AI to build new infrastructure from scratch in new terrain, in just a few years?

RG: There are definitely lessons in the use of real-time data and predictive intelligence. In a construction environment, for example, you can imagine data being captured on site safety, worker qualifications, equipment usage, weather conditions, materials, and supply chain movement. That creates a much richer picture of risk than many organizations have had before.

For insurers, the important point is not the construction example itself. It is what becomes possible when AI can connect live data to decisions. That could affect liability coverage, risk prevention, claims handling, and the way insurers understand exposure over time.

Catastrophe events are a good example. Insurers have traditionally responded after the loss occurs. AI creates the opportunity to act earlier. If a major storm or flood event is approaching, predictive intelligence can help identify which properties are most exposed and where intervention may reduce damage. In some cases, it may be better for an insurer to support mitigation before the event than to pay a larger claim afterward.

That is a different model. It moves insurance from being purely reactive to becoming more predictive and preventative. It is better for the insurer, and it is better for the customer.

IE: That really is the value of an AI operating system, always thinking ahead on weather, flood patterns and more, thinking of the best way to protect life and property. That’s the exciting opportunity, isn’t it?

RG: It is. The industry is moving from talking about AI to proving what AI can do. We are at the “show me” moment. Insurers want to see AI move from knowledge to action, and they want to see it do that within the realities of the insurance sector: regulation, governance, customer trust, and measurable business performance.

That is where AIOS comes in. It is not AI as a standalone tool or a talking point. It is insurance-native AI orchestration designed to help insurers put intelligence into the decisions that shape outcomes across pricing, underwriting, claims, customer engagement, and retention.

Ultimately, AI has to show up in the economics and operations of the business. It has to help insurers make better decisions, move faster, control risk, improve workflows, and create value for customers. It also has to do that within a governed framework, where insurers can explain decisions, audit them, and decide when human judgment should be involved.

Earnix saw an opportunity to help the industry move AI from experimentation into production. The focus now is on making AI work at scale, in the places where it can have the greatest impact.

IE: An exciting future ahead. It’s been interesting learning more, thanks for your time.