Tom Fairey, chief growth officer at Unlikely.ai, speaks about his firm’s entry for the Excellence in Claims Technology award
1. Why did you put yourselves forward for the Excellence in Claims Technology category?
Insurers are facing a real trust issue with current AI technologies. Many tools rely on probabilistic models such as large language models, especially in high-stakes claims decisions, which can be inaccurate or inconsistent and lead to severe consequences.
Our neurosymbolic approach offers a safer and more reliable alternative by automating claim decisions with high precision, consistency and transparency. We want to show that insurers can benefit from AI without lowering standards of accuracy or accountability. This award provides an opportunity to highlight a more trustworthy way of using AI in claims.
2. What do you think makes your entry stand out and why should you win the award?
Our neurosymbolic approach combines neural models with symbolic rules to deliver accuracy and consistency that large language models alone cannot provide. Neural components extract structured information, while symbolic programmes apply explicit constraints, check edge cases and generate clear justifications.
The system answers only when it is at least 99% certain and provides a consistent yes, no or don’t-know response. Ambiguous claims are automatically flagged for human review, enabling safe, controlled automation at scale. With major insurtech partners, it has delivered higher accuracy, faster processing and about 1.7 times more claims handled per person.
3. What would winning this award mean to you and your firm?
Winning the Excellence in Claims Technology award would recognise the real impact of our work in tackling one of the industry’s toughest challenges - trustworthy, scalable claims automation with AI.
It validates our belief that AI should be explainable and precise. For the team, it is a moment of pride and proof that building with integrity pays off. For clients and partners, it provides a third-party endorsement that UnlikelyAI delivers real-world value. For the wider industry, it signals that trustworthy AI is not a future goal – it is already here and it works.
No comments yet