In this April themed feature IE looks at how data has become an avalanche each day, third party platforms, plus the impact of AI when it comes to extracting maximum value.
Let’s get into it;
THINK ARCHITECTURE, THINK AI DESIGN, SAYS FEDERATO
William Steenbergen, co-founder and CTO of Federato, offers thoughts on AI structures, rather than trying to bolt AI tech onto legacy systems;
“Companies can’t get value from their data unless they can access it. P&C insurers working with old core legacy systems have been seeing this firsthand.
Buying AI tools and bolting them on fails. The lesson is that data architecture is destiny. There’s a big difference between AI-ready data and data isolated in silos or generic lakes.
When data isn’t AI-ready, progress dead-ends. The data is typically too slow or low quality to be useful, if it’s available at all.
The data an agentic AI system needs might be in the system somewhere, but it’s not a complete answer stored in a way that’s quick and easy to access. Instead it’s puzzle pieces scattered across outdated workflows, buried in antique code, and hidden behind inconsistent business rules.
The result is death by 1,000 pings. The system has to ping API after API to try to put the puzzle together. This is one reason AI pilots break down in production. The time for every task balloons. It just doesn’t work.
So when executives look at getting value out of data, they need to focus on the foundation. They need to ask, “can we access the right data at the right speed and quality at the right moment? And if not, what needs to change to enable that?”
Getting the best value from your data is urgent, so expect to see more investments in core systems that are AI native and consider data access for AI agents to be a core value driver.”
ALPHA FMC SEES MATURE TRANSFORMATION STRATEGIES
Britton Van Dalen, Global Head of Insurance Consulting at Alpha FMC said the industry is entering a more mature phase of transformation: ‘Insurers have been investing heavily in tech modernisation and digital capabilities over the last few years. In this new financial year, the focus shifts to operationalisation – ensuring those investments deliver real, measurable improvements in underwriting performance, capital efficiency and customer experience. Shareholder pressure – and client demand – is expediting this transition.
‘The insurers that manage to combine disciplined risk management with scalable technology platforms and strong data governance will be best positioned to differentiate themselves in a more competitive and volatile market.’
INFORMATICA OFFERS A DEEP DIVE
Levent Ergin, Chief Industry Strategist for Agentic AI, Regulatory Compliance & Sustainability, takes a deep dive into data issues and progress made so far by brands;
Insurers are investing heavily in AI, but many are building on data foundations that were never designed to support it.
Underwriting, pricing, claims, fraud detection – they’re all core insurance processes where AI can have a big impact. But in many organisations, the data behind these systems is fragmented, inconsistently defined, difficult to trace… It makes AI’s promise harder to realise in practice.
The root of the issue is how AI operates. It can’t perceive the real world; instead, it relies on interpreting patterns in data. If that data is incomplete, poorly governed, or doesn’t provide enough context, the AI’s outcomes will reflect those weaknesses – potentially at scale. For insurers, that signals underwriting exposure, higher costs, and questions to answer around regulatory compliance.
The move towards autonomous agentic AI-based systems could amplify these concerns. Now, instead of simply analysing data, the AI is tasked with acting on it – making decisions about underwriting, customer interactions, claims processing and more.
Think about claims, for example. Insurers are already deploying AI agents to analyse historical claims data, identify patterns, and flag potential fraud in real time. Others are introducing digital agents that guide customers through the claims process or provide real-time quotes, helping improve service while managing operational workload.
The benefits are clear for both insurers and customers – but again, a lot rests on the quality, governance, and context behind the data. Get it wrong, and the outcomes for either party could be unjustified and significant.
So, what can insurers do to make sure their data is fit for AI-enhanced operations – and to maximise the systems’ benefits? We’ve split it into three priorities:
Governance: not just compliance but differentiation
Strong data governance isn’t just a compliance exercise. Yes, it’s essential – especially in the light of tightening regulations. But, more than that, it can be a source of differentiation.
Insurers that can show clear data lineage, consistent definitions, and robust controls are well placed to move on AI quickly and with confidence. They can launch new products faster, respond to market shifts, and meet reporting obligations without relying on manual workarounds.
In many organisations, though, data remains fragmented. Information is duplicated, stored in different formats, and often disconnected from the processes that rely on it. That leaves teams piecing together datasets, checking outputs, and recreating reports – all slowing down their decision-making, and limiting their ability to scale with speed and confidence.
The solution can’t just be policy-based. It requires rethinking how the business runs to consistently embed data governance throughout the day-to-day. Consider: how are systems talking to one another? How does data flow between them? Do you have visibility?
The good news is, get data governance right, and transparency is a welcomed byproduct. And that’s important, as expectations grow on organisations to show which decisions we made using AI, how they were made, and what data informed them. With AI acting more autonomously, taking on a greater role in decision-making, insurers need to be able to explain and justify those outcomes with confidence.
Quality: the foundation of trust
How complete and reliable is your data? Is it of high enough quality that you’d trust an AI system to base a claims decision on it?
If the answer is ‘no’, then the likelihood is you have work to do across data accuracy, consistency, completeness, and currency. Remember: the results that an AI system produces can only ever be as reliable as the data foundation you provide it.
Consider a vehicle claims scenario. Insurers would typically base their decisions on things like telematics, repair histories, and customer profiles. But if that info is missing for certain regions, or it varies in quality, the model might misclassify legitimate claims as suspicious. That could have a knock-on effect on payout times, operational costs, and customer experience.
Data quality isn’t a quick fix. It needs continuous monitoring, standardisation, and clear ownership to keep issues identified and resolved early, especially as more decisions become automated.
Context: the reassurance of reliability
Even accurate data can lead to the wrong conclusions if it lacks context. Without it, AI systems can still produce outputs that seem logical but don’t reflect real-world conditions.
For insurers, that can lead to flawed decisions. Risk may be mispriced, customer behaviour misinterpreted, or emerging issues overlooked because the data is lacking elements that are crucial for correct interpretation.
Context enables data to be understood in practice. It includes where it came from, how it’s changed over time, and how it relates to other datasets. And it incorporates the business rules and assumptions that shape how it’s used. This is where metadata and lineage become critical, helping organisations understand how data is connected and how it informs decisions.
Insurers need to be able to understand and challenge AI’s outputs, meaning workforces must be trained on AI. We’re seeing confidence in AI growing faster than employees’ ability to fully interpret the data and models behind it. Closing this gap requires stronger data context, as well as greater data and AI literacy across the business.
The risk of getting it wrong
AI in insurance isn’t on the horizon, it’s already here, and it’s growing. With agentic systems, the reliance on AI to make the right decisions can’t be underestimated. That’s why ensuring the data foundations that insurers build their systems on must be bulletproof. Flawed data, at best, could lead to mispriced risk, incorrect claims decisions, and inconsistencies in how cases are handled. Worst case, though, those issues compound, affecting customer trust, financial performance, and regulatory compliance.
There’s much at stake, but if insurers address data governance, quality, and context as priorities, their confidence with AI can grow exponentially.
Be the first to comment