Only 14% of insurance firms have fully integrated AI into their operations. That number sits at the center of a new survey covering 250 managers across the UK and US — and it tells a more complicated story than a simple technology gap.
According to the report, published by AutoRek, the obstacle is not reluctance. 82% of firms expect AI to dominate the industry. The problem is what lies underneath: fragmented data, legacy systems, and manual processes that corrode operational budgets before any automation gets a chance to run.
The firms surveyed spend 14% of their operational budgets correcting manual errors. Nearly half operate settlement cycles longer than 60 days. Around 22% of respondents identify reconciliation complexity as a significant driver of cost increases, and a similar share link existing inefficiencies directly to governance and audit risks.
Seventeen Sources, One Problem
The average firm in the survey manages 17 distinct data sources. A majority flag this as a problem, and the report notes it compounds significantly after mergers and acquisitions, where data estates from multiple organizations collide without clean integration. Governance frameworks built on top of fragmented data become fragmented themselves. The authors describe this as the primary reason AI deployments remain constrained across the sector — not the AI itself, but the architecture it would have to sit on.
Six percent of companies report no use of AI at all.
Transaction volumes, the report projects, will rise by roughly 29% over the next two years. Without structural changes to how firms process and reconcile data, operating expenditure is expected to climb in step — driven by the same combination of manual processing, siloed systems, and transactional complexity that already burdens the sector. The authors note that their previous publications raised these same structural issues, and that the findings now entering the public domain are not new warnings.
Where AI Can Start
Rather than a wholesale transformation, the report points toward reconciliation as the logical entry point for AI deployment. It is a bounded, rules-based domain where automation can generate measurable results quickly, making it a viable testing ground before firms attempt broader integration. The authors are direct about the limits, though: any automation layer placed over a fractured data architecture will struggle to scale without driving costs higher, not lower.
On the structural data problem, the report suggests cloud-based AI platforms as a more practical path than in-house builds, citing their potential to help firms consolidate and structure fragmented sources. The implication is that data consolidation and AI adoption are not sequential steps — firms that treat them as separate workstreams may find neither delivers what the 82% are expecting.
Photo by Pixabay
This article is a curated summary based on third-party sources. Source: Read the original article
