How AI is Reshaping Commercial Insurance and Risk Assessment – with Sidharth Ojha of AXA XL

Commercial insurance coverage has lengthy struggled to undertake new expertise on the tempo of different monetary providers. Manual workflows, outdated mainframes, and fragmented techniques from years of mergers have slowed modernization efforts. Many insurers nonetheless view underwriting as an “artwork” moderately than a course of, which has traditionally delayed even primary digital upgrades.

Industry information underscores the substantial adoption hole throughout the insurance coverage sector and past. In MIT Center for Information Systems Research’s world examine of enterprise AI maturity, solely 7% of organizations have absolutely embedded AI throughout operations, whereas most stay in pilot or mid-stage phases. At the identical time, regulatory agendas are lastly catching up. 

The EU AI Act got here into impact in 2025, requiring insurers to categorize AI techniques by danger degree and comply with strict transparency guidelines. Meanwhile, the overwhelming majority of enterprise information — more than 90% — is unstructured, saved in paperwork, contracts, and PDFs which might be troublesome to research with out superior instruments.

This combine of legacy techniques, compliance calls for, and information challenges creates a essential inflection level for insurers. How can they undertake AI responsibly whereas guaranteeing ROI and minimizing danger? Drawing on insights from Sidharth Ojha, Head of Process Optimization, Data & AI at AXA XL, in a current episode of the AI in Business podcast, this text explores how business insurers can modernize operations, empower groups to experiment, and lay the foundations for scaling.

This article examines three key insights from Ojha’s perspective on AI adoption in insurance coverage:

• Empowering enterprise customers with low-code AI: Provide underwriters a compliant sandbox to experiment safely and uncover constraints early.
• Turning information right into a strategic asset: Map information finish to finish and convert unstructured contracts into structured insights that drive development.
• Building foundations for scalable AI: Standardize roles, processes, and information definitions to forestall pilots from stalling and unlock enterprise adoption.

Listen to the complete episode under:

Guest: Sidharth Ojha, Head of Process Optimization, Data & AI, Global Chief Underwriting Office, AXA XL.

Expertise: Commercial Insurance Transformation, Process Optimization, and Applied AI

Brief Recognition: At AXA XL, Ojha leads initiatives to use AI in underwriting and operations, balancing compliance with effectivity and cultural change. His expertise spans legacy course of modernization, regulatory alignment, and enabling sensible AI adoption in a single of the world’s largest business insurers.

Empowering Business Users with Low-Code AI 

Ojha sees that, among the many clearest challenges for driving AI adoption in insurance coverage, is cultural inertia. Executives typically acknowledge AI’s potential however hesitate to let non-technical employees interact with it instantly, which Ojha sees as a missed alternative.

He describes the significance of creating “protected lanes” the place underwriters and enterprise customers can take a look at AI instruments in managed environments. By embedding low-code platforms into present techniques, insurers can allow experimentation with out risking information leaks or regulatory breaches.

“Think of it like bowling with bumpers,” Ojha explains. “You wish to let individuals take the shot, however hold them from rolling into the gutter.” His strategy builds confidence and helps uncover limitations early, earlier than a mission absorbs vital finances or time.

In the previous, insurance coverage tech initiatives relied on prolonged handoffs: enterprise analysts translated necessities, builders constructed techniques, and architects ensured alignment. By the time options reached manufacturing, essential context was typically misplaced. Low-code AI instruments allow underwriters to work together with expertise instantly, bypassing translation layers and accelerating actionable suggestions.

Ojha stresses that leaders mustn’t rush to pilots or MVPs. Instead, they need to allocate extra time to exploration and failure within the sandbox part.

“The extra time you spend failing your hypotheses, the much less time you waste scaling one thing that doesn’t work,” he notes. For an business the place “failure” carries damaging connotations, reframing the necessity for failure tolerance as managed testing may help insurers undertake AI extra comfortably.

This cultural shift is important for adoption. By giving underwriters direct however safeguarded entry, organizations create buy-in and align instruments with actual enterprise wants — moderately than constructing in isolation and hoping for adoption later.

Turning Data right into a Strategic Asset 

Ojha insists – as many earlier podcast company have – that expertise alone can not ship ROI with out clear, usable information. He notes that Insurance firms face a very steep problem as a result of most of their essential data is locked in unstructured codecs, resembling coverage paperwork, endorsements, quotes, and schedules of values.

Ojha factors out that 5 years in the past, insurers struggled to do one thing as primary as studying a desk in a PDF. Generative AI has solved many of these hurdles, however unstructured information stays various and inconsistent, making transformation into structured codecs troublesome:

“Most of the information insurers depend on isn’t even of their techniques — it’s trapped in PDFs, Word paperwork, and scanned contracts. The actual problem isn’t studying it, it’s standardizing it. Each coverage is distinctive, typically written like a authorized manuscript. Until we are able to constantly flip that unstructured information into structured data, each downstream AI use case — from danger evaluation to pricing — will likely be working at midnight.”

— Sidharth Ojha, Head of Process Optimization, Data & AI, AXA XL

The payoff is vital. With structured information, insurers can reply portfolio questions in seconds, resembling: “Which insurance policies exclude communicable illness?” or “How a lot publicity do we have now throughout a area?”

During the COVID-19 pandemic, many organizations couldn’t reply shortly to such queries. Today, AI instruments provide the possibility to keep away from that blind spot.

Ojha additionally describes new prospects in summarization capabilities amongst these techniques. Beyond condensing paperwork, he notes that AI can examine shopper submissions in opposition to inner urge for food and compliance guidelines. 

For high-volume underwriting groups, these capabilities imply touching extra submissions per day, declining unsuitable dangers sooner, and specializing in worthwhile alternatives. “That’s not simply effectivity,” Ojha stresses. “That’s actual development potential.”

For leaders, the mandate is clear: deal with information as a first-class asset. Inventory coverage wordings, goal high-volume ache factors, and construct techniques that push structured outputs again into core platforms. Done effectively, these steps rework AI from a cost-saving software right into a income driver.

Building Foundations for Scalable AI 

While pilots are acquainted with insurance coverage, scaling stays uncommon. Ojha estimates that “80-90%” of AI initiatives stall between proof of idea and deployment. The causes are much less about expertise and extra about organizational readiness.

He outlines the information infrastructure bottlenecks that always derail scaling AI operations in insurance coverage:

• Unclear accountability for information fields, resulting in inconsistent inputs.
• Fragmented processes, the place groups report completely different ranges of element for a similar product.
• Legacy stacks which might be costly to combine with new AI fashions.
• Inconsistent definitions of key metrics throughout enterprise items.

Without fixing these foundations, even promising pilots fail to increase. Ojha advises leaders to ask: If this resolution went dwell throughout three nations tomorrow, what would break first? Addressing gaps in that framework upfront prevents pricey surprises later.

Regulation additionally performs a job, and Ojha sees the EU AI Act as a turning level, offering classes that boards and regulators alike can belief. 

“If you might be compliant with EU guidelines, you might be largely compliant globally,” he notes, insisting that having such assurance can ease govt considerations and speed up mission approvals.

Ultimately, success comes from endurance. Insurers are sometimes keen to leap from concept to MVP, however Ojha emphasizes the worth of deeper exploration and testing. Companies that spend money on readability of roles, course of alignment, and information high quality will discover it simpler to maneuver AI from experimentation to enterprise-wide adoption.