AI Doesn't Replace Digital Transformation. It Makes It More Valuable.

Why Digital Transformation Matters More Than Ever in the Age of AI

Why Digital Transformation Matters More Than Ever in the Age of AI

There is a growing belief across the insurance industry that AI agents will replace traditional underwriting systems.

The thinking goes something like this: why spend time and money building underwriting platforms when an AI agent can simply understand the process, make decisions and interact directly with users?

It's an attractive vision. Unfortunately, it is also the wrong place to start.

More importantly, it risks distracting insurers from the work that will determine whether they succeed with AI.

AI has made digital transformation more urgent, not less. Insurers that continue to delay modernising their pricing, underwriting and operational processes will find it increasingly difficult to realise the full value of AI. The organisations moving fastest to digitise today will be the ones best positioned to compete tomorrow.

The digital-first misconception

Many insurers still rely on spreadsheets, emails and manual workflows for underwriting, pricing and operations.

When they decide to modernise, they often see two apparent choices:

1. Build a traditional hard-coded underwriting platform.
2. Skip the platform entirely and let AI do the work.

Neither option is ideal.

Traditional systems can become expensive to build and maintain, with functionality embedded deep in code. Users often struggle with adoption because every system behaves differently and requires training.

On the other hand, a pure AI approach expects the AI to understand every nuance of underwriting, governance, regulation and company-specific processes. Today's AI models are remarkably capable, but they still perform best when operating within a well-defined business framework.

The better approach: Digital first, AI second

The real opportunity is to separate the business process from the AI.

First, create a robust digital platform that captures data, manages workflows, applies governance and provides the operational backbone that underwriters already understand.

Then connect AI to that platform.

The AI doesn't replace the underwriting system, it becomes an intelligent interface to it.

Users can ask questions in natural language, retrieve information instantly, automate repetitive tasks and receive recommendations, while the underlying platform continues to provide structure, governance and auditability.

More importantly, AI starts helping underwriters do what they do best. Rather than spending time navigating systems, searching for information or completing repetitive administration, underwriters can focus on assessing risk, making decisions and handling more business. AI augments human expertise instead of attempting to replace it.

AI becomes dramatically cheaper

This approach has another significant advantage: cost.

If every business rule, calculation and workflow is delegated to a large language model, organisations quickly discover that token consumption becomes expensive.

Instead, the digital platform performs the deterministic work it was designed for. AI is only used where it genuinely adds value, reasoning, summarisation, explanation, search and user interaction.

This fundamentally changes the economics of AI. Rather than asking an LLM to perform every calculation and execute every workflow, organisations use AI where it creates the greatest value while allowing the platform to perform the structured processing it was designed for. The result is significantly lower operating costs and a far more scalable AI strategy.

Databricks has spoken extensively about the importance of AI operating alongside structured systems rather than replacing them entirely. Well-designed AI "harnesses" reduce unnecessary context and dramatically improve efficiency. The less an AI model has to rediscover each time, the fewer tokens it consumes, the better it performs, and the more commercially viable AI becomes at enterprise scale.

This isn't just an architectural improvement, it becomes a commercial one.

The learning curve almost disappears

One of the biggest barriers to digital transformation has always been user adoption.

• People need training
• They need manuals
• They need support

When AI is connected to the platform through technologies such as Model Context Protocol (MCP), the learning experience changes completely.

Instead of learning menus and documentation, users simply ask:

• "How do I create a new quote?"
• "Why was this referral generated?"
• "Show me similar policies."
• "Which pricing model should I use?"

The AI understands both the application and the business context, dramatically reducing the learning curve.

The result is software that becomes easier to learn, easier to use and more intuitive for everyone, regardless of technical expertise. Instead of adapting to software, users simply interact with it in everyday language.

The second-order effect

The most exciting benefit appears once multiple systems become connected.

Historically, information has lived in silos. Data exists, but users rarely know where to find it.

With AI connected across digital platforms, every piece of information becomes discoverable. The AI can locate the relevant data, combine insights from multiple systems and present answers in the context of the user's task. People stop searching for information. Instead, they apply judgement. That is where experienced underwriters, actuaries and executives create value.

A practical example

We've seen this approach ourselves. Rather than attempting to build an AI-native underwriting platform from scratch, we first built a digital underwriting platform with structured workflows, governance and business logic. We then connected it to an MCP server.

Immediately, users could interact with the platform using natural language.

Nothing about the underlying business process had to change, but the experience became dramatically simpler.

Even more importantly, most of the heavy lifting continued to be performed by the platform itself, at a fraction of the cost of asking an AI model to perform every calculation and workflow.

The result wasn't simply a more intelligent platform. It became easier for underwriters to use, easier to train new staff, faster to navigate and significantly cheaper to operate.

The AI became an intelligent layer over an efficient digital platform.

The future isn't AI first

The future isn't AI replacing digital systems.

Insurers thatinvest in robust digital foundations today will be in the strongest position toexploit AI tomorrow.

The winnersover the next five years won't be the organisations that attempt to buildAI-first insurance systems from scratch. They'll be the insurers that builddigital-first, AI-enabled platforms, where AI enhances existing workflows,empowers underwriters and unlocks greater value from structured data.

Digitaltransformation has not become less important because AI has arrived.

It has becomemore important than ever.

The message forinsurers is clear. Digitise first. Connect AI second.

That's how organisations will achieve faster adoption, lower operating costs, better underwriting decisions and ultimately realise the full value of AI.