• Simon Axon, Global Financial Services Strategist at Teradata


• 11.03.2026 11:30 am
#AutonomousAI #AIBanking

Last year, agentic AI entered the lives of both consumers and businesses raising everyone’s expectations for a more autonomous way of working. The agents were considered an indispensable tool that reasons, plans, and acts around the clock but they need to be grounded in enterprise knowledge to properly improve customer service as well as employee efficiency. 

The banking sector has already started adopting agentic AI into their operations - for example,  preparing client materials, modelling financial scenarios, and initiating proactive outreach. As more businesses embrace the technology, we’ll see a lot more banks experiment and apply the tool more widely within their strategies.

Agentic AI’s Capabilities

Agentic AI is able to independently plan, make decisions, and immediately take action. In contrast to other automation tools that follow a fixed path, agents don’t rely on pre-determined instructions. These autonomous tools use learnt patterns to decide on the next, best move that would assist them with achieving their objectives.

Agentic AI automates manual processes giving back valuable time for employees to concentrate on more creative and urgent projects. The technology boosts efficiency while it continues to learn and adapt to improve the business’ workloads. Having said that, for agentic AI to work, it needs a huge amount of data to perform the analytics required for effective action. 

Not only that but context plays a huge role in ensuring the results are trusted and in line with the bank’s goals. This helps reduce hallucinations and increase reliability in AI agent responses. To make this happen, businesses should build a knowledge foundation through context engineering so the tools can navigate any complexities and act autonomously. 

Context engineering addresses a critical challenge where AI models trained on general knowledge lack understanding of the specific business environment. When personalised to your sector and aligned with your business, these agents learn your objectives, priorities and challenges and work accordingly.  

Last year, we saw practical applications moving from theory to production. For example, generative AI (GenAI) copilots and autonomous portfolio agents turned to multimodal AI that can analyse text, voice, and transaction data for real-time fraud detection.

How Banks are Benefitting

The banking sector is also leveraging agentic AI and has already seen some tangible benefits. For example, during the Google Cloud event last year, a challenger bank in the UK realised impressive results, including a 40% reduction in onboarding time and a 30% reduction in false positives in anti-money laundering.

Similarly, at the Evident AI Symposium in New York City last October, Mastercard mentioned that they use the tool on fraud prevention at the point of sale. They specifically add more context to transaction data and build a merchant graph for each cardholder to spot suspicious patterns. The business has successfully integrated this merchant-graph signal into its fraud models to successfully confirm cardholder intent and minimise fraud risk.

It’s worth mentioning though that true AI at scale is only secured when it’s properly embedded into the core processes so that it improves everyday moments for customers.  Not only that but without a proper foundation, banks risk receiving outcomes that are not reliable. 

The Era of Hyper-Personalisation

Agentic AI is also used for swiftly creating bespoke solutions to customers so much so that products are hyper-personalised in real time. At the Data Lab’s session on scalable AI in Edinburgh last year, panelists described a future of financial inclusion, where a bank could create tailored protection products on the spot for a customer with a £50 monthly budget.

As we look into 2026 and beyond, hyper-personalised financial agents will be able to manage an individual’s financial life, from tax optimisation to debt consolidation. 

Watch out for Regulation

There’s no doubt of the competitive advantage banks can gain from agentic AI, but we shouldn’t neglect the importance of remaining compliant. Looking at the EU AI Act, agentic AI will be considered as high-risk meaning that it’ll need to be traceable, auditable, and have human oversight at all times. The EU AI Act will impose its most stringent obligations by August 2026 reshaping how AI is adopted and governed across the EU. 

Regulation was also a key topic during Davos. Day two saw leaders discuss how regulators should work with the financial sector to ensure they maintain stability and foster innovation. The conversation also highlighted the importance in engaging in a multistakeholder collaboration with technology partners which would be critical considering the important role agentic AI can play for the banking sector.

Banks that use compliant agentic AI will achieve quicker client onboarding and enhance productivity but they will also see an increased client engagement and be considered a trusted partner.

Banking-as-a-Service

Along with agentic AI, banking-as-a-service (BaaS) architecture is another important element supporting the banking sector. BaaS lets banks monetise their regulated capabilities by embedding payments, accounts, lending and compliance services directly into third‑party digital journeys, extending reach beyond traditional channels.

Success in BaaS depends on having access to industrial‑grade data and strong risk and compliance foundations, enabling banks to expose services safely while meeting stringent audit, lineage, and security requirements.

When it comes to delivering BaaS capabilities, scalability and predictable economics are essential, with platforms needing to absorb volatile partner demand through high‑concurrency processing, mixed‑workload management and fixed‑cost models with optional elastic burst.

Having a unified data fabric also accelerates partner onboarding, reducing integration effort by providing a single, governed source of truth that supports real‑time decisions, open table formats and virtualised views.

Banks differentiate by powering their partners with AI-enabled intelligence, from fraud signals to personalised insights, using modern architectures that support LLMs, vector search and agent‑based automation. 

But without scalable infrastructure, unified data, and AI differentiation, the architecture remains a high-touch, low-margin business that can't deliver on its promise of seamless financial services integration.

The Future of Banking

We are at an exciting technological stage where agentic AI and BaaS are able to significantly transform the banking sector and continue evolving as they constantly learn from it. 

The real-life agentic applications we’ve seen so far have demonstrated the technology’s true potentials and how it can be used to not only cut down on manual processes but also protect the business from illicit activities and support customers with bespoke solutions.

But banks will not be able to produce trusted and unbiased results without closely following the ever-evolving regulatory requirements. It’s therefore critical to remain compliant for the business to stay competitive and reliable.