Financial technology provider SBS has launched the SBS AI Foundation, a sovereign artificial intelligence platform designed to bring generative AI and, eventually, agentic AI into core banking environments without requiring sensitive banking data to leave an institution’s controlled infrastructure. Powered by NVIDIA technology, the platform supports private cloud, managed cloud and on-premises deployments. SBS says the system is already running in production at Scottish Building Society in the UK on Amazon Web Services, with availability extended to its wider financial institution customer base.
SBS Brings Sovereign AI Infrastructure to Core Banking
SBS has introduced an AI platform designed to help banks deploy generative AI while retaining control over sensitive customer, transaction and operational data. The SBS AI Foundation combines banking software with NVIDIA’s AI infrastructure, targeting financial institutions that want to automate operations and improve decision-making without compromising data governance.
The launch addresses a central challenge in banking technology: applying advanced AI to regulated financial workflows while meeting requirements for privacy, security, explainability and data residency. SBS says the platform can run in private cloud, managed cloud or fully on-premises environments, allowing institutions to align deployment with their infrastructure and regulatory needs.
The company, which says more than 1,500 financial institutions rely on its technology, has already deployed the platform in production at Scottish Building Society, a long-standing customer in the UK. SBS reports that the implementation has delivered approximately 30–40% faster response times with more consistent outputs. The announcement does not specify the workloads, baseline measurements or testing methodology behind those figures.
Bringing AI Closer to Banking Data
Banks hold sensitive information across core banking, lending, digital banking and regulatory reporting systems. Using that information with external AI services can create challenges involving data transfers, third-party processing, security controls and jurisdiction-specific requirements.
SBS AI Foundation is designed to make information across its banking infrastructure available to generative AI applications while keeping deployment within the institution’s controlled environment. According to SBS, the platform uses NVIDIA NIM microservices to power the AI layer, NVIDIA NeMo Guardrails to help constrain model responses and open Nemotron models as part of its technology stack.
The architecture aims to support AI applications grounded in a bank’s own data rather than relying exclusively on general-purpose model knowledge. That distinction matters in financial services, where answers about products, customer accounts, lending or compliance need to reflect institution-specific information and operating rules.
However, sovereign deployment does not automatically guarantee regulatory compliance or eliminate every security risk. Banks still need appropriate access controls, model evaluation, data protection measures, audit processes and oversight of AI-generated outputs. The extent to which the platform satisfies specific obligations will depend on its configuration and the institution’s implementation.
Three Initial Areas of Application
SBS identifies three main use cases for its platform.
Customer engagement and personalization: Generative AI can use a bank’s records and product information to support onboarding and customer interactions. The goal is to help institutions identify relevant products and improve operational efficiency while making AI outputs more explainable.
Natural-language data intelligence: A generative AI copilot is designed to translate questions about structured and unstructured data into usable insights. This could help product teams and executives retrieve information without waiting for lengthy reporting processes or specialist analyst support.
Risk and compliance intelligence: AI-powered monitoring is intended to identify potential risk and compliance exposure while producing an audit trail. Keeping this processing within the bank’s controlled environment may help institutions manage sensitive information, although the effectiveness of risk detection and the completeness of audit records require validation in practice.
SBS says these capabilities can support faster decisions, product development and operational workflows. It has not disclosed customer adoption figures or quantified financial savings associated with each use case.
Agentic AI Is the Next Planned Step
SBS plans to extend the platform toward agentic AI, in which autonomous agents execute multi-step banking workflows with human oversight. Potential applications include processes that require information gathering, decision support and coordinated actions across banking systems.
Moving from generative AI assistants to agents introduces additional operational considerations. An agent that can act on banking systems needs clearly defined permissions, identity controls, transaction authorization, monitoring and safeguards against inappropriate or unintended actions.
SBS says future agentic capabilities will operate within a governed environment with human oversight. The company has not announced a general availability date for this planned functionality.
The distinction between current and planned capabilities is important: the existing platform is described as live in production, while the expansion into autonomous workflow execution remains a future development.
NVIDIA Partnership Targets Regulated AI Deployment
The platform combines SBS’s banking technology with NVIDIA’s AI stack, including NIM, NeMo Guardrails and Nemotron models. NVIDIA described the approach as a shared intelligence layer beneath SBS’s banking platform, intended to support AI across cloud and sovereign infrastructure.
SBS also cited a Gartner statistic indicating that 84% of finance organizations had implemented or planned to implement AI, while only 7% reported high or very high impact from those investments. These figures, attributed to Gartner in the announcement, highlight the gap between AI adoption and measurable business results; the underlying report and survey methodology should be checked before using the figures as a broader industry benchmark.
For banks, the commercial question is whether sovereign AI can move beyond experimentation into dependable, auditable production systems. SBS is positioning its platform as infrastructure for that transition, combining access to banking data with deployment flexibility and governance controls.
Its early production deployment provides an initial reference point, but broader adoption will depend on independently validated outcomes, integration complexity, security performance and the ability to maintain effective human oversight. As financial institutions evaluate AI for core operations, the balance between useful automation and controlled execution is likely to remain a defining requirement.
Market Landscape
Banking AI is moving from isolated pilots toward deployment in operational systems, where data security, auditability and regulatory accountability are essential. SBS AI Foundation targets this transition by combining generative AI capabilities with banking infrastructure and deployment options designed for institutions with strict data residency requirements.
Three market dynamics stand out:
- Sovereign AI infrastructure: Banks increasingly need deployment architectures that give them control over where sensitive information is stored and processed.
- Governed enterprise AI: Guardrails, explainability, access controls and audit trails are important for applying AI to lending, reporting, customer engagement and risk management.
- Agentic banking: Autonomous agents could eventually coordinate multi-step workflows, but financial institutions will need strong permissioning, transaction controls and human oversight.
SBS’s production deployment at Scottish Building Society provides an early example of its approach. The reported 30–40% improvement in response times is a company-reported result, not yet a broadly validated industry benchmark. Adoption at scale will depend on measurable operational benefits, security assessments and integration with existing banking systems.
Top Insights
- SBS AI Foundation combines generative AI technology with banking infrastructure, allowing institutions to deploy AI within their controlled environments.
- The platform uses NVIDIA NIM microservices, NeMo Guardrails and open Nemotron models to support governed AI applications.
- SBS reports 30–40% faster response times and more consistent outputs in its production deployment at Scottish Building Society.
- Initial applications include customer engagement, natural-language data analysis, and risk and compliance intelligence with audit trails.
- Agentic AI for multi-step banking workflows remains a planned expansion, with human oversight and governance central to the proposed approach.
Get in touch with our fintech expert


