With support from the Innovate UK BridgeAI programme through a Smart Grant, Holistic AI completed Phase 1 – an initial feasibility study – to explore the development of an AI Risk Management Platform tailored to the insurance industry. Building on this foundation, the company has now completed Phase 2 under the UKRI Accelerating Trustworthy AI programme (valued at approximately £1.12 million), advancing the platform from concept into delivery and strengthening the evidence base for commercialisation.

Challenge

AI is increasingly embedded in critical decision-making processes, creating challenges for insurers to ensure systems are safe, fair and compliant. Many organisations currently lack consistent ways to evaluate and mitigate the reputational, legal and commercial risks associated with AI.

Unlike established risk domains such as credit, operational or cyber risk, there are currently no widely adopted tools that quantitatively assess AI-specific risks across technical, ethical and regulatory dimensions. Traditional GRC systems were not designed to assess issues such as bias, model robustness, explainability, privacy leakage or performance drift at scale.

As a result, insurers face a significant gap between growing AI adoption and their ability to evidence control, assurance and compliance. This challenge is compounded by limited visibility into informal or decentralised AI usage (“shadow AI”), which can bypass governance processes altogether. Holistic AI is changing this, aiming to support enterprises to adopt and scale AI responsibly by providing visibility, risk monitoring, auditability and policy enforcement across AI systems.

Solution

Holistic AI’s Phase 1 feasibility study focused on designing an AI Risk Management Platform specifically for the insurance sector. The proposed platform evaluates and quantifies AI-related risks and supports organisations to manage the development and deployment of AI systems through auditing tools and pathways to insurance.

The concept builds on Holistic AI’s existing GRC software, which audits and assures AI systems across code, data, policies and processes. Holistic AI also enables organisations to identify and catalogue “shadow AI” by discovering AI tools and models in use across the business, including those not formally registered or centrally tracked, helping teams build a more complete view of their AI landscape within the insurance sector.

The Holistic AI platform brings together four core capabilities:

• Discovering AI in use to automatically find and list AI systems across an organisation, including internally built tools, third-party software and embedded AI features. This creates a single, up-to-date picture of where AI is being used.
• Assessing risk by evaluating across key areas such as fairness, reliability, transparency and data privacy. The platform then produces a risk score that helps organisations understand which systems need the most attention.
• Ongoing monitoring to continuously check for changes in behaviour, performance issues or policy breaches, so new risks can be spotted early.
• Supporting action and reporting by guiding teams on how to reduce risk and producing reports that support internal governance and regulatory compliance.

With Phase 2 now completed under the UKRI Accelerating Trustworthy AI programme, Holistic AI progressed from feasibility into platform development and validation. This phase focused on translating user requirements into a deployable product pathway, strengthening sector-specific functionality, expanding technical capability for risk quantification and monitoring, and deepening alignment with insurance governance and compliance needs. Phase 2 also supported closer engagement with industry stakeholders to test assumptions, refine workflows and ensure the platform can integrate with insurer operating models.

With Phase 2 now completed under the UKRI Accelerating Trustworthy AI programme, Holistic AI progressed from feasibility into platform development and validation. This phase focused on translating user requirements into a deployable product pathway, strengthening sector-specific functionality, expanding technical capability for risk quantification and monitoring, and deepening alignment with insurance governance and compliance needs. Phase 2 also supported closer engagement with industry stakeholders to test assumptions, refine workflows and ensure the platform can integrate with insurer operating models.

Impact

Phase 1 enabled Holistic AI to form a consortium and validate demand for a dedicated AI risk platform in insurance, a sector where AI adoption has been comparatively cautious. Innovate UK support also enabled Holistic AI to develop an initial sector-focused report that captured priority risk areas, user requirements and technical capabilities. This report now acts as a reference point for product direction and the next phase of development.

With Phase 2 completed, Holistic AI has moved beyond concept validation to demonstrate credible delivery momentum, strengthening readiness for commercialisation and adoption. The solution is expected to help organisations of different sizes increase confidence in AI use while reducing exposure to risk, aligning with national priorities around trustworthy and responsible AI. The platform is designed to support alignment with internal policies and external regulatory frameworks, helping enforce compliance.

Future

With Phases 1 and 2 complete, Holistic AI is well-positioned to commercialise and scale the platform, aligned with the recommendations and sector insights generated through BridgeAI engagement and subsequent development work. The company plans to expand its offering across the insurance sector and into other regulated industries, supporting organisations to adopt AI with confidence and compliance.

In September 2023, Holistic AI submitted formal written evidence to the House of Lords Communications and Digital Committee’s inquiry on large language models, sharing its insights on the development, risks and governance of LLMs to help inform the Committee’s report and future policy considerations.

AI risk can’t be managed like traditional risk domains if you can’t see what’s running, how it behaves, or how it changes over time. Our work with BridgeAI helped validate the need for an insurance-focused AI Risk Management Platform that scores key risks like security, robustness, bias or transparency, and turns that insight into audit-ready action.

– Adriano Koshiyama, Co-Founder Holistic AI