Innovations in artificial intelligence (AI) and data are transforming the way financial services are delivered, with Scotland demonstrating what is possible.
Applications of advanced technologies in finance are enhancing the quality of services provided to customers and optimising the efficiencies of internal operations for businesses.
Scottish startups and spinouts are rapidly developing into successful businesses with considerable influence in the UK and beyond. For all the talk of future uses, the innovations in AI and data are delivering value today.
Download sample pages of selected reports
Explore a selection of report samples we have handpicked for you. Get a preview of the insights inside. Download your free copy today.
How AI is delivering value in finance in Scotland
There has been much hype around AI and its multiple potential applications. But the technology is now making a day-to-day difference through fintech in Scotland.
Based in Edinburgh, Aveni is a specialist provider of AI fintech solutions in domain-specific large language models (LLMs) and advanced natural language processing (NLP). The company began as a University of Edinburgh startup, and its solutions are now used by some of the most-recognised names in finance.
Aveni’s solutions have been engineered to automate complex processes and enhance productivity in the intricate financial services workflows. Solutions have also been developed to deliver regulatory compliance and data-driven insights for continuous improvements.
The company is developing LLMs specifically for financial services by enhancing its existing products, Aveni Assist and Aveni Detect. Fast, accurate analysis is a core feature of Aveni Detect, which currently relies on purpose-built LLMs that are now being refined.
Identifying customer vulnerabilities in the transcripts of long calls more quickly could potentially decrease the amount of time staff spend reviewing calls in risk and compliance by as much as 30%-50% a month, as well as improve decision-making and reduce risk for staff.
Aveni Assist has been designed to optimise the workflow of advisors and boost response time while reducing the time spent on administration. The solution is also being refined and benchmarked on various LLM models.
Aveni Detect and Aveni Assist have been deployed by the customer-facing service team from Octopus Money, with each client interaction now analysed automatically. This has led to an 85% increase in call visibility, from 15% to 100%. Time savings enable teams to focus more on improving quality and skills development.
In another example provided by Aveni, Prosser Knowles had put automation at the core of its digital transformation strategy. The company uses Aveni Assist to streamline workflows for advisers, reduce manual administration, and enhance the quality of client services. Crucially, Aveni Assist was introduced to existing systems rather than introducing new platforms.
The solution records interactions with clients, summarises meetings, drafts reports for customer suitability, and delivers compliance with regulations. Advisers can spend more time improving the quality of their advice and building engagement with clients, saving an estimated 60% of time on post-meeting admin.
The evolution of LLMs in financial services
AI is evolving. The initial wave of enterprise AI relied on extensive and generalised LLMs to appear fluent, but models were not trained for specific uses in financial services. With accuracy a critical component in financial services and explainability a regulatory requirement, any AI model deployed must understand the industry. Given that language use notoriously changes from industry to industry, there are obvious flaws with the blanket use of generalised systems.
But there are signs of change. This year, research from Nvidia emphasised the greater efficiency and lower running costs of small language models (SLMs) for agentic systems in specific industries. Researchers found that SLMs also offer greater flexibility and control, pointing to a future where AI agents are smaller and smarter in more specialist applications.
These findings are also evident in Aveni’s FinLLM product, released in May. Built specifically for UK financial services, FinLLM is a domain-specific large language model (LLM). In contrast with generalised models, FinLLM has been designed to comply with FCA guidance and the EU AI Act. The system can perform at scale while being accountable and trustworthy.
FinLLM has been trained to function optimally in the language, context, and nuances of financial services. Smaller models can also be deployed to handle specific datasets and functions. This reduces the reliance on third-party application programming interfaces (APIs) to provide greater security for sensitive processes.
Developed over the last year and driven by advanced NLP research, FinLLM was supported by investment from Nationwide and Lloyds Banking Group. Such collaborations enabled FinLLM to be developed in accordance with real-world finance industry use cases and regulatory frameworks.
“In an era where AI sovereignty is becoming increasingly important, FinLLM is a fantastic example of UK AI Innovation. It combines the brilliant minds in Aveni Labs, many of whom come from the University of Edinburgh and makes Aveni one of the strongest AI Labs in the country. The result is a highly performant model that is delivering proven automation for a range of use cases in UK Financial Services,” Jamie Hunter, COO of Aveni.
“Additionally, we have a strong investment market in Scotland and recognition that commercial innovation and academic excellence, when best able to work together, deliver transformational outcomes,” adds Hunter.
Synthetic data explained
When building new products, software systems, or AI tools, banks are limited by what customer data they can use. Furthermore, banks operating across borders must often follow different regulations, such as GDPR in Europe and CCPA in California.
“Data is incredibly personal and private. It’s really strongly controlled, quite rightly,” says David Tracy, head of data product at bigspark. The company was listed in The Times’ Top 100 Fastest Growing Private Companies in 2024.
For international projects, issues obtaining data can cause major delays. A solution has been developed by bigspark’s Scottish-based team, Aizle, which provides an innovation known as synthetic data. The company can set specific parameters to generate high volumes of simulated data, which could include monthly income and outgoings, direct debits to pay bills, mortgage payments, and other personal spending habits.
“We simulate all of that in these really large-scale AI-driven simulations,” explains Tracy. “When we run those to generate the data, it can be used for various cases. For instance, in open banking, payments, insurance, and energy.”
Different levels of accuracy are possible in the synthetic data depending on the project’s use cases. Tracy explains how, as with any type of AI or machine learning, the accuracy is dependent on how much time can be spent on the process.
“For the accuracy or quality of synthetic data, there are three different dimensions to it. One is the fidelity – how closely does it resemble the real world?” says Tracy
“The second is privacy. Our method solves this completely, but in some of the other methods, if you’ve used real-world data to create new data from that through some kind of AI process, how well is the privacy of the original people protected in the data?
“And the third is what we call utility, which is how useful it is. With our method, we’ve made it completely privacy safe. So, there’s never any risk there. But that means that then we need to really focus on making it really useful and high-quality. That comes down to the use case and what you need the data to do for you.”
Simulated data for finance projects with reduced risks
With financial information an increasing target for sophisticated levels of cybercrime, using synthetic data can help to reduce the risks involved with data sharing.
Bigspark’s Aizle software builds datasets based on an understanding of the original data. If working with a bank, the partner will be able to demonstrate what data looks like from a customer, an account, or a transaction, and then create benchmarks to measure it against. This information can then be used to scale up synthetic datasets.
Banks have significant volumes of data, but there are issues with security and permissions. Bigspark worked with a large bank that wanted to experiment with AI tools but was conscious about compliance with regulations. Bigspark provided a solution in synthetic data to test the AI framework.
“They were able to test and contrast all these methods and figure out what was useful for them in a kind of hackathon-type environment,” adds Tracy.
At the other end of the business scale, one of the main issues for fintech startups is data availability. For example, they may have an idea for a financial product to help people manage their money, but lack the data to begin the project. Bigspark can provide data to allow the startup to build its first prototype and customer demo without having the issues of data scarcity or relying on data from its staff, family, or friends.
The company has also been working with the UK Financial Conduct Authority (FCA) on a project to combat fraud. Synthetic data has been provided for a collaboration across academia, fintech, banking, law enforcement, and government regulators.
“It’s to really understand how this data can help explore big questions about how data sharing in these law enforcement-type situations should work between banks, the police, and others,” says Tracy. “The problem is you can’t use real data for that because the frameworks don’t exist yet. So, we’ve created the synthetic data to allow that innovation to happen, to allow that exploration really safely, with cross-sector collaboration between companies and organisations big and small.”
Another notable project for bigspark has been with the UK Department for Business and Trade to create a world-leading synthetic dataset that imagines what the UK’s data infrastructure might look like in ten or 20 years. More smart data schemes could soon follow, with the UK Government recently passing the Data Use and Access Act to provide the necessary legal powers domestically.
The environment supporting fintech growth in Scotland
Both bigspark and Aveni have their roots in projects at the University of Edinburgh. The innovative solutions provided to a growing client base would not be possible without the supportive environment to develop and grow. The University of Edinburgh has an impressive track record of producing startups and spinouts in fintech, notably through the Smart Data Foundry, which Aizle was originally part of.
Looking ahead, a recent software breakthrough by researchers at EPCC, the UK’s National Supercomputing Centre at the University of Edinburgh, could enable future AI models to process information ten times faster than existing systems.
Fintechs based in Scotland or considering being based there also have potential access to a significant client base and market opportunities. Central to the industry ecosystem is FinTech Scotland, which provides support to fintech businesses of all sizes. The organisation can facilitate connections to further innovations through collaboration. In addition, the Data Lab is another positive force within the fintech cluster in Scotland.
As the economic development agency of the Scottish Government, Scottish Enterprise also plays a pivotal role in supporting fintech businesses to grow in Scotland. Through the agency, companies either based in Scotland or interested in investing in the country can access a broad range of support, including funding, grants, and advice. Grants are available to create jobs and innovate, as well as for capital expenditure.
The government agency recently introduced a new type of capital grant to support capital investment with transformative potential in Scotland. Alongside this, Scottish Enterprise manages a series of co-investment funds targeted at early-stage companies, in addition to the Scottish Loan Scheme.
Furthermore, Scotland is an internationally recognised centre for financial services, with major banks based in the country regularly collaborating on projects with innovative fintechs.
“We’ve got really talented people here in Scotland. We want to build on that. We’ve got great universities in Glasgow, Edinburgh, and St Andrews in that triangle in the Central Belt. So, we have lots of super intelligent grads coming out of there, but then there are professionals who are more experienced in financial services,” adds Tracy. “Obviously, Edinburgh is well-established as a world centre for financial services. But Glasgow’s hot on the heels.”
To learn more about investing in Scotland and the opportunities available, download the document below.
Unlock up to 35% savings on GlobalData reports
Use the code at checkout in the report store
-
20% OFF
Buy 2 reports
Use code:
Bundle20
-
25% OFF
Buy 3 reports
Use code:
Bundle25
-
30% OFF
Buy 4 reports
Use code:
Bundle30
-
35% OFF
Buy 5+ reports
Use code:
Bundle35
Valid on all reports priced $995 and above. Cannot be combined with other offers.
Still deciding what will work best for your business?
Ask our experts for help.
Enquire before buying
