Travel Insurance
2 Nov 2026|Mandy Langfield
Featured in
ITIJ 310 | November 2026
Share
AI claims processing viability depends on the technology, the investment, and appetite of the insurer, and with deployment rolling out ever faster across the industry, Mandy Langfield takes the opportunity to speak to experts to dive deeper into what success looks like
Regulated industries adopt technology the same way every time: slowly, then comprehensively. The pace at which investment is happening is accelerating almost by the day in 2026. A recent blog from SimplAI highlighted key industry trends worthy of further exploration.
Insurance, by its very nature, has always been document-heavy, workflow-heavy, labour-intensive, and regulated. All of this slows the adoption of new technology. However, it also means opportunity for growth, investment, and development. The more manual work there is to automate, the bigger and better the application of artificial intelligence (AI) will be.
Various insurers around the world are reporting massive gains through their use of AI in different lines of insurance.
Aviva has deployed more than 80 AI models in claims, and said it cut liability assessment time by 23 days, improved correct routing by 30%, and reduced complaints by 65%, resulting in savings of over £60 million in 12 months.
Allianz built Project Nemo, a seven-agent claims system for its business in Australia, and has since reported an 80% reduction in claim processing and settlement time. According to a report from Sixfold, Zurich North America saved up to two hours per claim submission across more than 200 underwriters.
Reality bytes in travel
Within travel insurance, the implementation of AI into the claims process is happening faster than expected, according to industry experts ITIJ spoke to for this article.
Carl Carter of CSC Global Partners reported that he is working with “multiple travel insurers, MGUs, and TPAs that are introducing AI into their travel insurance claims process”. Why is it that travel insurance and AI claims fit so neatly together? High volume, low value – think travel delay, lost or damaged luggage. And travel is seasonal, so insurers need to be able to manage surge periods, which, when you need to hire a dozen new claims handlers every summer and train them, gets expensive.
What do customers get when their insurer uses AI for claims processing? Typically, they would benefit from faster settlements and a better experience. For the insurer, lower leakage rates, better fraud detection (more on that in the next article!), and the ability to free up your staff to focus on more complex claims are all obvious benefits.
The blocker has always been the need for a human to review incoming documentation to check it is valid and matches what the customer has declared
Other drivers pushing AI claims processing forward, said David Corney, CEO of Firemelon, are pressure on claims handling costs, and the difficulty of recruiting quality staff. He went on to point out: “The blocker has always been the need for a human to review incoming documentation to check it is valid and matches what the customer has declared. However, this is where AI is really changing things, as it can read these documents and carry out the checks which previously had to be manual. I am expecting this to unlock a lot more claims that can be handled without any human intervention over the next 12 months.”
Develop your own or not?
Is there a right answer when it comes to the issue of whether or not you should develop in-house or buy off the shelf? Size matters.
“Larger insurers are focusing on in-house innovation and AI teams, along with the implementation of chief AI officer roles in the C-suite,” Carter told ITIJ. “MGUs and multi-line capacity-focused insurers tend to be buying in AI subject matter expert services by working with AI-led digital native claims handlers and TPAs that are not encumbered by legacy platforms.”
Identifying where AI can provide the most value in the claims process is key. Industry experts report that claims triage and routing are where AI has incredible power and influence. Determining claim type, urgency, and complexity means that the claim is routed to medical specialists, fraud experts, and human handlers as needed.
Pavel Gertsberg is CEO at ClaimSorted, a digital-first AI-native UK, EU and US multi-line claims handler and TPA. He explained the company’s role succinctly: “We apply AI to the operational layer of claims, not the decision layer: automated FNOL ingestion and data entry (around 98% of data entry is automated via portal and policy admin integration), document summarisation and chasing, AI-assisted fraud scoring on every claim, day-one reserve estimation, and an AI voice agent that handles around 95% of non-assistance medical calls.”
It is not just the pay and file claims that are benefiting from AI, but also assistance case handling and complex medical claims too
While the industry once said AI claims would only be used for the simplest claims, that no longer seems to be the case. Carter told ITIJ: “It is not just the pay and file claims that are benefiting from AI, but also assistance case handling and complex medical claims too.”
Medical claim management
The use of automated document ingestion and understanding to extract and validate invoices, boarding passes, bookings, police reports, and baggage documentation removes manual data entry and is therefore a huge time saver. As Carter points out, complicated and emotive medical claims are not currently in the running for AI claims management, and quite rightly. He said that there is a limit to the kinds of cases that AI is equipped to handle: “Such cases would include repatriation of mortal remains, death, and PA claims – as well as claims where there are complex pre-existing medical condition disclosures or omissions.”
This is an opinion with which Arlenee Lebrón Segura, Head of Fraud and P&C Claims Manager in AXA International Markets, concurs wholeheartedly: “I believe human judgement, expertise, and empathy should always remain central whenever decisions have a significant impact on people’s health, wellbeing, or trust. In health insurance and assistance, this includes serious illnesses, highly complex treatments, hospitalisations, medical evacuations, and repatriations. In these situations, AI should augment human judgement, not replace it.”
I believe human judgement, expertise, and empathy should always remain central whenever decisions have a significant impact on people’s health, wellbeing, or trust
Gertsburg holds the same point of view, and the ClaimsSorted system is clear that coverage determinations and complaints handling are always human; AI provides a day-one reserve estimate. Still, the decision belongs to the handler, and complaints sit with a human team. Complex or high-net-worth claims stay human-led (dedicated calls, no automated journeys), and vulnerability cases. “Minors, fatalities, and distressed customers are escalated by the AI voice agent to a person immediately,” he told ITIJ.
There are clearly aspects of travel and health insurance claims where AI will almost certainly never be able to make its mark. Segura makes this abundantly clear: “A specialised case manager supporting patients with serious illnesses does far more than coordinate benefits or authorisations. They help reduce uncertainty at one of the most difficult moments in a person’s life. They explain what will happen next, guide patients on how to access the care they need, and support them throughout the entire journey so they understand what to expect from their insurer and how they will be supported, regardless of the complexity of their cover. When dealing with serious illness, reducing uncertainty is almost as important as reducing processing times. Technology can accelerate processes, but it cannot replace the reassurance, professional judgement, and empathy that experienced professionals provide.”
Statistics demonstrate success
Carter reported: “AI processing success rates depend a lot on the technology, investment, and appetite of the insurer (or if they use a digital-first AI-native claims service). I have seen ranges from 25% through to 70%, and potentially higher by volume (not value). From a human manual handling perspective, this equates to between 50–90% reduction in manual claims handling time.”
For ClaimSorted, around 70% of the work within a claim is automated, and low-value, low-fraud-score claims (typically sub-£500–£1,000, depending on client-set thresholds) can flow straight through with a single human approval touch point. All claims retain a human in the loop by default – AI flags, a handler approves. On the specialist side, 10% or more of claims are manually QA-reviewed, large losses (£25,000–£30,000-plus) get team lead review every 14 days (30 in litigation), and flagged fraud escalates to the company’s internal special investigations unit (SIU).
The success of AI should not be measured by the percentage of claims that are automated, but by its ability to enable professionals to devote more time to delivering value where it truly makes a difference for policyholders
For Segura, however, data can vary so widely between insurers using AI at this relatively early stage in adoption that we should focus more on other metrics: “The success of AI should not be measured by the percentage of claims that are automated, but by its ability to enable professionals to devote more time to delivering value where it truly makes a difference for policyholders.”
She added: “Alongside traditional indicators such as claims cycle time, Net Promoter Score (NPS), and customer satisfaction, I believe it is equally important to incorporate metrics that assess the quality of AI-assisted decisions. These include the rate of human review of automated decisions, false positive and false negative rates in fraud detection, the override rate of AI-generated recommendations, and the number of times policyholders are asked to provide the same information during the claims journey.”
Keeping the human element
Customers must be made aware that they are dealing with an AI platform when they come to claim. And are they happy with this status quo? NPS scores would suggest this is the case – customer satisfaction improves, claims handling time reduces, and payment receipts go out faster. Gertsberg reports that NPS runs at 60-plus on the firm’s strongest programmes, and complaint rates sat at 0.41% for Q2 2026.
More importantly, the issue of when the human needs to step in for the insurer has to be clear. Most insurers ITIJ spoke to for these articles were keen to highlight that what AI is doing is actually enhancing the roles of their staff members, as they are able to focus on the more complex aspects of their job and not be weighed down by data management.
Carter wholeheartedly agrees when it comes to claims: “Most important is an AI decision and feedback log, where the ‘human in the loop’ can review and audit these decisions, calculations, and interactions.”
Automate the simple and repetitive; keep human expertise wherever empathy, judgement, or coverage interpretation drives the outcome
Corney was keen to point out that as an industry, we don’t yet know for sure how much of the insurance claims process can be accurately automated. He told ITIJ: “Clients have rightly automated the easiest, lowest-value, and low-risk claim types first, and over time will widen those rules to pick up more complicated claims. However, how far that will go into seriously complicated claims – who knows?”
November 2026
Issue
In this, our biggest issue ever, our main focus is on artificial intelligence. After recent threats and warnings about the potential dangers of AI and calls for coordinated action to slow the technology’s development, we speak to experts about the technology in a three-part special. We look at data-driven risk assessment, underwriting, and portfolio optimisation, which are all being transformed by AI, and ask what insurers and underwriters are doing to get the most out of the technology.
Read full issue
Travel Insurance
2 Nov 2026
Share
Mandy Langfield
Mandy Langfield is a Director of Palladium Media Partners and former Publishing Director at Voyageur Group. She has written extensively on international travel and health insurance, medical assistance provision, and air medical transportation.
