UnlikelyAI Teams Up With SBS Insurance Services

It’s always good to test out AI by deploying it in real time, here’s the word; 

UnlikelyAI, the company pioneering low-risk enterprise AI by combining neural networks with symbolic reasoning, reveals the results of its first deployment in regulated industry after years of stealth development. Partnering with InsurTech firm SBS Insurance Services, the pilot successfully automated 40% of claims handling with 99% accuracy, and provided a full audit trail for 100% of decisions, delivering enhanced customer service and setting a new industry standard.

This new use case addresses the biggest barriers to AI adoption in regulated industries like insurance, which must comply with standards including transparency and explainability of decision-making. Until now, LLMs have fallen short due to flaws in their decision-making leading to inconsistent claim outcomes and lengthy processing times. Through UnlikelyAI’s solution, SBS has achieved accurate, controllable and transparent decisions.

Despite perceived efficiencies, traditional large language models (LLMs) present unique challenges for regulated industries, as evidenced by UnlikelyAI’s research:

• Hallucinations: Even advanced LLMs like GPT-4.5  produce incorrect or misleading responses in up to 37% of factual questions, according to OpenAI’s own analysis. In insurance, this means extensive human oversight is required, delaying customer claim resolution.

• Inconsistent outcomes: Despite the need for uniformity of claim outcomes for a fair and compliant service, LLMs consistently produce variable outputs. UnlikelyAI’s research showed different models, such as Deepseek R1 and Meta’s Llama, marking anywhere from 8% to 46% of identical claims as ‘covered’. And roughly one in ten times, no matter which LLM was used, the same platform generated a different response for the same claim.

• Auditability: Regulations require explainable decisions but LLMs are ‘black boxes’ with inaccurate, or non-existent, explanations of how they produced their outputs.

SBS selected UnlikelyAI because its unique approach combines traditional neural networks with symbolic reasoning – a transparent form of AI that follows strict rules. This enables SBS to fully explain decision-making to customers and regulators with complete transparency and precision in a scalable way.

UnlikelyAI’s software complements SBS’s existing expertise, achieving 99% accuracy (compared to ~52% for LLMs) while deferring to human judgement where appropriate. This allowed SBS to automate 40% of test claims without hallucination concerns and reduce miscategorisation by 99% – enabling faster, more consistent customer service with a human touch.

William Tunstall-Pedoe, Founder and CEO of UnlikelyAI, said: “We built UnlikelyAI to address the fundamental limitations of large language models, especially in high-stakes industries where trust, precision and transparency are essential. This deployment – our first to be made public – demonstrates that AI doesn’t have to be a black box; it can be transparent, explainable and genuinely accountable. Our approach enables regulated businesses to confidently adopt AI without compromising on compliance, consistency or customer trust.”

Bazil Crowley, Director of Innovation at SBS Insurance Services, added: “We’ve always prioritised excellent customer services and using technology as an enabler for that. UnlikelyAI’s neurosymbolic approach stood out as a game-changer for our operations, with this partnership allowing us to deliver a faster, more reliable service to our customers whilst maintaining our human touch where it matters most. We’re excited to lead the way for this milestone in InsurTech and are proud to be setting new industry standards for intelligent claims processing.”