Across insurance, there is no shortage of excitement about AI.
Across insurance, there is no shortage of excitement about AI.
Every board has it on the agenda. Every strategy presentation references automation, operational efficiency or intelligent decision-making. Yet behind the headlines, the reality inside many organisations is far more complex.
Drawing on in-depth interviews and research for the TINtech Data Jam agenda, one theme came through consistently from insurers, brokers, MGAs and technology leaders alike:
The challenge is no longer whether AI can deliver value. The challenge is how organisations adopt it successfully without creating new operational, regulatory and cultural risks.
The research revealed a sector that is simultaneously ambitious and cautious. There is growing confidence that AI can genuinely improve productivity, reduce turnaround times and remove repetitive administration. But there is also a growing recognition that many organisations are underestimating what it takes to move from pilot projects through to BAU that delivers business transformation.
The industry is no longer debating the opportunity
What stood out most was how pragmatic the conversation around AI has become.
This is no longer about experimentation for experimentation’s sake. The focus has shifted toward very practical business problems:
- Reducing manual administration
- Improving claims turnaround times
- Streamlining onboarding and compliance processes
- Enhancing portfolio oversight
- Extracting value from fragmented data
- Improving underwriting consistency
- Removing operational bottlenecks
One participant described AI not as a standalone technology initiative, but as “another layer of operational transformation.” That distinction matters.
The most mature organisations are no longer treating AI as a separate innovation programme sitting outside the business. Instead, they are embedding it into broader operating model discussions around workflows, customer experience, data quality and decision-making.
That shift in mindset is important because many of the barriers to AI adoption are not technical at all.
Most AI failures are not technology failures
One of the strongest themes to emerge was that insurers are often trying to solve problems in the wrong order.
Several participants reflected on projects where organisations attempted to automate highly complex processes before addressing underlying operational issues such as fragmented systems, inconsistent data or unclear workflows.
The result is predictable. Automation accelerates complexity rather than reducing it.
One participant summarised the issue clearly: “AI can automate parts of a process successfully, but organisations struggle when trying to automate end-to-end workflows. If one failure or hallucination breaks trust in the process, confidence disappears almost instantly.”
This reflects a wider challenge across the industry. Many insurers are still operating with legacy processes designed for manual intervention, disconnected systems and institutional knowledge held inside teams rather than structured data environments.
In many cases, AI is exposing operational weaknesses that already existed.
Data remains both the opportunity and the constraint
Another recurring theme was the industry’s continuing struggle with data complexity.
Insurance organisations have spent years discussing data transformation, yet many participants acknowledged that underwriting and operational decision-making still relies heavily on individual experience, informal judgement and inconsistent data structures.
One described underwriting decisions being driven by “80-plus parameters that only exist in people’s heads.”
That observation gets to the heart of the challenge.
AI performs best in environments where processes, workflows and data structures are already reasonably mature. But many insurance organisations are still dealing with:
- Inconsistent data capture
- Over-engineered reporting structures
- Multiple legacy platforms
- Manual workarounds
- Duplicate processes
- Market-specific complexity
- Regulatory fragmentation across territories
The research also highlighted growing frustration around excessive data requirements. Several participants questioned whether organisations are collecting far more information than they genuinely need.
This is becoming an important strategic question for insurers. The winners may not be the organisations with the most data, but the ones that define the minimum viable data needed to make effective underwriting and operational decisions.
Governance is becoming a major adoption bottleneck
While the industry often talks about the risks of AI itself, many organisations are discovering that their internal governance structures may be an equally significant barrier.
A number of respondents described situations where low-risk AI use cases were being forced through governance processes designed for the highest-risk scenarios. The result was lengthy approval cycles, excessive scrutiny and stalled implementation.
This is creating a growing tension inside large organisations.
On one hand, firms recognise the need for strong governance, explainability and regulatory oversight. On the other hand, overly rigid governance frameworks risk slowing adoption to the point where innovation becomes almost impossible.
The insurance market is now entering a phase where governance maturity may become as important as technical capability.
The organisations that progress fastest are likely to be those that develop proportionate governance models aligned to the actual level of risk involved.
The real challenge is cultural, not technical
Perhaps the most important insight from the research was the extent to which AI adoption is ultimately a people challenge.
Again and again, the conversation returned to fear, uncertainty and trust.
Employees are worried about job displacement. Leaders are worried about accountability. Underwriters are worried about losing professional judgement. Operational teams are worried about being forced into unfamiliar ways of working.
One of the strongest conclusions is that successful AI adoption depends heavily on psychological safety.
People need space to ask questions, challenge assumptions and admit uncertainty without feeling exposed. Organisations that frame AI purely as a cost reduction exercise risk creating resistance before adoption has even started.
Interestingly, several participants noted that enthusiasm for AI adoption did not always correlate with age or technical background. In many cases, experienced industry professionals were among the strongest advocates once they understood how AI could reduce repetitive work and improve decision-making.
That creates a powerful opportunity.
Used correctly, AI may help insurers retain and scale institutional knowledge that would otherwise leave the industry as experienced professionals retire.
Insurance may need to be bolder
One of the more thought-provoking comments was whether insurance has become too willing to accept complexity as an excuse for slow progress.
Compared with sectors such as banking, where real-time data exchange and automated decisioning are now standard, insurance still operates with significant operational friction across many workflows.
The tools increasingly exist. The appetite increasingly exists. The business case increasingly exists.
What remains unclear is who will lead the change and how quickly organisations are willing to rethink traditional operating models.
Four themes likely to shape the TINtech Data Jam agenda
The results of the research will shape conversations at TINtech Data Jam including:
- How insurers move from isolated AI pilots to enterprise-wide operational transformation
- The role of governance, explainability and trust in scaling AI safely
- Why data simplification may matter more than data expansion
- How leaders create the cultural conditions needed for successful adoption
The market has clearly moved beyond asking whether AI matters.
The more important question now is whether insurance organisations are prepared to redesign processes, governance models and operating cultures deeply enough to realise its potential.
Sign Up to TINsights
Where we share our latest blogs, industry reports and insights
