AI in Insurance 2025 – Why the Human Still Leads

Until recently, conversations about AI in insurance revolved around experimentation, running PoCs, exploring novel use cases, and automating specific steps. 

 Today, that’s changed. The question isn’t “Where could we use AI?”, it’s “Why do we need it?” 

 This mindset marks a shift from tech experimentation to business optimisation. AI is no longer “innovation theatre.” It’s delivering measurable value through speed, accuracy, profitability, and better customer experiences. 

AI That Helps, Not Replaces 

AI in insurance today is practical, and human led. 

Instead of replacing underwriters, it’s helping them: 

• Triage faster: Scoring submissions based on appetite and urgency 

• Price smarter: Supporting segmentation and loss forecasting 

• Automate claims and fraud detection: Reducing leakage and speeding up decisions 

• Review documents: Extracting clauses and summarising policies with LLMs 

• Streamline operations: Supporting teams via internal AI chat assistants 

In other words, AI is not replacing judgment, it’s removing friction. 

From Build vs. Buy to Configure & Integrate 

As insurers mature in their use of AI, they’re also shifting their implementation strategy. 

The old “build vs. buy” debate is being replaced by a more agile approach: configure and integrate. 

Instead of reinventing the wheel, teams are adopting modular platforms with flexible APIs, embedding AI directly into existing workflows, with faster time to value and minimal disruption. 

 Related: Build vs. Buy in Insurance →

Leading insurers are already seeing measurable improvements.

Hiscox, for example, cut underwriting turnaround from 72 hours to just 3 minutes in its sabotage and terrorism business line using AI. 

Meanwhile, McKinsey estimates that AI could generate up to $1.1 trillion in annual value for the global insurance industry.  

Still, many MGAs, especially small and mid-sized ones in the US and UK, haven’t yet realised that value. Beyond the digital-first MGAs with strong delegated authority, most are still navigating cost, complexity, and operational limitations. 

To gully capitalise on AI, they’ll need to move past technical barriers and adopt solutions that work with real-world insurance workflows. 

But What’s Holding AI Back? 

Despite its potential, AI still faces blockers: 

• Poor data quality 

• Siloed data across functions and departments 

• Fragmented legacy systems 

• Cultural resistance (“It’s just tech”) 

• Uncertainty around explainability and compliance 

• High implementation costs 

• A crowded and confusing vendor landscape 

Scaling AI isn’t just about tools. It’s about building trust, and that means clear communication, strong governance, and human-centred design. 

The Human Still Leads 

The real power of AI in 2025 lies in human-AI collaboration: tools that amplify human judgment, streamline manual work, and free up expert talent to focus on higher-value, strategic decisions. 

But real impact comes when teams feel in control, when AI is an enhancer, not a replacement. 

 At inari, we believe AI should give teams their time back, not take ownership away. 

 Want the Full Picture? 

This blog scratches the surface, and we’re not done yet. 

Stay tuned as we dive deeper in our upcoming piece: 

• Want to know which AI use cases are delivering real ROI? 

• Curious about how others are navigating regulation and compliance? 

• Wondering what’s actually slowing adoption down, and how to solve it? 

We’re unpacking it all soon. Make sure you don’t miss it.