Sign in to view
member only content

Spotted: There has been much recent discussion about the dangers of AI. For example, a recent survey of 2,800 executives by Deloitte found that around 75 per cent feel their organisation is not prepared for the emerging risks that AI presents. Working to develop trust in AI, Armilla AI provides risk mitigation and transfer solutions to help empower enterprises to safely deploy AI tools.

Armilla provides proprietary evaluation technology that assesses and quantifies the risk level of AI models. The company provides a holistic evaluation that analyses data quality, model performance and accuracy, fairness, interpretability, and robustness, identifying safe, ethical, and trustworthy AI that is in line with regulations and standards.

By evaluating models, Armilla helps AI vendors detect hidden biases that can impact protected populations, identify areas of weakness in their models, and establish use-case-specific thresholds for acceptable behaviour of the technology. This analysis, in turn, helps vendors build credibility and future-proof their technology.

With backing from leading insurers Chaucer, Greenlight Re, and Swiss Re, the company provides an Armilla Guaranteed warranty for AI models that are found to be safe and effective when measured against key metrics. This means that if a model does underperform, companies that have bought the AI tool can receive a refund on the product, helping to de-risk the purchase of these third-party technologies.

Armilla is a graduate of Y-Combinator, Lloyd’s Lab, and Betaworks’ AI Camp accelerators. The company recently raised a $4.5 million (around €4.1 million) seed funding round led by Mistral Venture Partners, with participation from insurance providers Greenlight Re and Chaucer as well as MS&AD Ventures, SixThirty Ventures, Morgan Creek Digital, and Y Combinator.

With AI being used for increasing numbers of applications, including discovering mineral deposits and streamlining nature restoration, it is vital that organisations are able to trust the models that they use. Armilla could be instrumental in providing this reassurance.

Written By: Lisa Magloff