Summary
- The FCA's AI Spotlight Observatory covers examples from more than 200 organisations across 150 areas of financial services activity.
- CFTE helped develop its taxonomy and research structure, while NayaOne designed and built the platform.
- The public resource distinguishes use cases from actual implementations; registration is required for selected detailed insights and demonstration requests.
The Financial Conduct Authority has launched an online observatory containing examples of artificial intelligence applications across financial services, giving organisations a structured way to examine where technology is being used and how different approaches compare. The AI Spotlight Observatory draws on material from more than 200 organisations spanning 150 areas of activity.
Introduced on 19 November, the resource has been developed with the Centre for Finance, Technology and Entrepreneurship, or CFTE, and technology partner NayaOne. It provides a common framework for categorising AI use across financial services, distinguishing proposed applications from technologies that have been implemented in practice.
The Observatory covers activities ranging from financial crime detection and compliance to emerging uses such as agentic payments and neurosymbolic AI. Its public interface is available without charge, while registration is required for more detailed information, including selected target customer data and the ability to request demonstrations from participating innovators.
The FCA describes the platform as the first of its kind introduced by a financial regulator. Its purpose is to make practical examples more accessible rather than evaluate or approve every AI product included within the resource.
Common definitions may improve technology comparisons
Financial services organisations increasingly encounter applications described using broad AI terminology, making it difficult to distinguish between systems that perform similar tasks and those offering substantially different capabilities.
A technology supplier might describe a product as an intelligent assistant, an AI agent or an automated analytical service, even where its practical function overlaps with applications marketed under another label. Such differences can complicate the early stages of technology research and procurement.
The Observatory introduces a common taxonomy covering sectors, business functions, workflows and the value that applications are intended to create. That structure is designed to make use cases easier to compare without relying exclusively on the terminology chosen by individual vendors.
It also distinguishes between a potential use case and an implementation. The difference is important because identifying a plausible application does not demonstrate that a financial institution has deployed it successfully or obtained measurable commercial benefits.
An AI system intended to detect suspicious transactions, for example, may be at an early stage of testing or already incorporated into operational risk processes. Understanding its maturity helps organisations assess whether the technology has been demonstrated under conditions comparable to their own requirements.
However, categorisation cannot by itself establish that an implementation is effective. Details about accuracy, operational performance, cost and governance may still be needed before a business can determine whether an approach is suitable.
More than 200 organisations contribute examples
The FCA says the initial Observatory includes use cases from over 200 organisations across 150 areas of activity. These examples provide a wider view of the technologies being explored by financial services businesses, although the database should not be treated as a representative statistical survey of the entire sector.
Companies contributing use cases may have greater interest in demonstrating their technology than organisations that do not participate. The resulting collection can be useful for identifying available products and developments without establishing the prevalence or effectiveness of every application.
CFTE supported the development of the taxonomy and the organisation of the underlying repository, while NayaOne was responsible for its design, navigation, search and filtering capabilities.
The platform builds on the FCA’s previous AI Spotlight Digital Showcase, adding a more consistent framework for comparison and additional ways to identify patterns across the market. New cases are expected to be incorporated as organisations submit examples for consideration.
That continuing development may make the resource more useful over time, particularly where emerging technologies are difficult to categorise through existing definitions. Nevertheless, changes to the dataset will need to be understood when making comparisons between different periods.
AI adoption is widespread but uneven
The Observatory arrives as AI adoption becomes increasingly common in UK financial services. A separate Bank of England and FCA survey published in October 2026 found that 86% of responding firms were using AI, up from 75% in the corresponding 2024 survey.
Those findings also showed substantial differences in the types of technology deployed. Generative AI accounted for 56% of reported AI systems, while the use of automated agents was widespread but generally involved limited autonomy.
Among responding firms, 80% reported using AI agents in production to provide decision support, although only 2% reported agents with broad autonomy. These results illustrate why categorisation by practical function and operational maturity is useful when assessing adoption.
An organisation using AI to recommend actions while employees retain authority to make decisions faces different implementation requirements from one allowing automated systems to execute transactions or modify operational controls independently.
The survey also found growing dependence on external AI suppliers. Third-party implementations accounted for 39% of reported systems, increasing from 33% in 2024, while the concentration of computing and hosting relationships raised additional questions about technology dependencies.
Such findings provide context for the Observatory, but the two sources serve different purposes. The survey measures adoption among respondents, while the Observatory assembles documented examples to help organisations examine technologies and their possible applications.
Greater visibility does not replace due diligence
The regulator’s initiative could help smaller organisations discover technologies relevant to established business problems without undertaking the same breadth of initial research independently. It may also assist researchers examining how AI is being introduced into financial operations.
Access to information about available applications does not, however, establish whether a system is suitable for a particular financial institution. Customers must still assess data security, performance, regulatory obligations, integration requirements and the consequences of errors.
For regulated firms, the decision to use AI does not remove accountability for customer outcomes or the management of operational risk. An externally supplied model may form part of a process, but the firm remains responsible for meeting the obligations associated with the service it provides.
Comparisons become particularly demanding when applications use different datasets, evaluation methods or definitions of successful performance. A tool that performs well in one environment may produce different results when applied to another institution’s information and operating procedures.
Implementation costs also matter. Software can require substantial work to connect with established systems, establish appropriate controls and prepare employees to use it responsibly. Those costs may influence the value of an application as much as the initial price of accessing its technology.
The Observatory complements existing regulatory programmes
The FCA has developed several mechanisms for examining AI, including its AI Lab, Supercharged Sandbox and AI Live Testing programme. These activities support engagement and controlled experimentation, while the new Observatory provides broader access to examples and market information.
There is an important distinction between participation in an innovation programme and regulatory endorsement. Inclusion in the Observatory should not be interpreted as evidence that the FCA has approved an application, certified its performance or determined that it meets every regulatory requirement.
The regulator is also pursuing work with the Monetary Authority of Singapore to support AI innovation and collaboration across the two markets. The Observatory provides a potential reference point for that cooperation, alongside other initiatives addressing the testing and deployment of financial technology.
As more organisations contribute examples, the resource may provide a clearer picture of which types of AI applications are becoming available and how their functionality is changing. Its usefulness will depend on the quality of the information supplied, the consistency of its classification framework and the evidence available about actual implementation.
The Observatory expands access to practical market intelligence, but the distinction between a demonstrated use case and a commercially reliable deployment remains central. Financial firms can use the information to identify potential technologies; determining whether those systems deliver suitable outcomes will still require their own evaluation.












