Financial services firms entered 2026 determined to move both generative and agentic AI applications decisively from experimentation to enterprise deployment. Organizations are spending heavily on pilots, proofs of concept, and innovation programs built around AI-assisted customer service, internal productivity, software development, fraud detection, and compliance. This investment must deliver real business value and client-centric outcomes.
The Business Challenge: Moving From Pilot Programs to Real ROI
Generative and agentic AI have become the two dominant areas of investment across financial services; many firms see them as integral to the next wave of productivity, automation, and customer experience. Gen AI can accelerate research, content creation, and service interactions, while agentic AI promises to automate more complex workflows and decisions. Today’s challenge is turning these investments into enterprise-scale value and avoiding AI sprawl: too many pilots, too many overlapping tools, and too many isolated projects that never scale beyond the experimentation phase. Without a clear, holistic, firm-wide strategy, firms risk creating what many technology leaders now describe as a pilot graveyard. This means beginning with clearly defined business objectives tied to customer experience, workflow acceleration, operational efficiency, or employee productivity.
Keys to Successful Implementation and Scaling
Success increasingly depends on establishing clear risk frameworks and governance processes. Financial services firms already have mature model risk management practices for predictive AI, and many are now extending those same principles to generative models. This includes testing for bias, documenting training data, monitoring outputs, validating performance, and creating controls for human oversight. Financial institutions cannot afford to let AI operate unchecked in areas where decisions can affect credit access, fraud claims, investment outcomes, or erode customer trust.
Choosing the right use case is the foundation for any AI implementation. For gen AI, many firms are experimenting with hyper-personalization use cases to deliver value, because their customers are demanding more bespoke products and services. When it comes to agentic AI, rather than deploying autonomous agents everywhere at once, firms need to identify areas where more advanced orchestration genuinely adds value beyond traditional automation. As Financial Technology Today has previously explored, financial institutions are already beginning to apply AI agents to areas such as compliance, where automation can reduce repetitive work while allowing human investigators to focus on higher-value decisions.
In many cases, optimizing existing automation capabilities first creates a stronger operational foundation for more sophisticated agentic deployments over time. Many firms are finding that software development and IT operations provide a strong starting point. AI agents can support coding, testing, quality assurance, security reviews, and infrastructure management. These are environments where outcomes are easier to measure, risks are more contained, and organizations can learn how to manage agent behavior before deploying it into more sensitive customer-facing processes. Financial Technology Today has also examined the importance of starting small when overcoming barriers to enterprise AI adoption.
In both cases, firms are recognizing the need for a centralized AI platform that gives them visibility into all models, all workloads, and deployments across the enterprise. With a common platform across clouds or on-premise environments, AI becomes just another workload. When you abstract away complexity and standardize at the lowest common denominator, organizations can improve resilience and focus on security and innovation instead of managing complexity, helping reduce AI sprawl. Red Hat’s financial services platform strategy similarly focuses on helping financial institutions scale AI across hybrid environments while addressing security, compliance, cost, and operational resilience.
This also gives operations teams a single control plane to manage their infrastructure across all Kubernetes environments. If this is not done at the infrastructure layer, you will have silos that waste resources—which is compounded when you add in AI workloads. Institutions seeing stronger ROI are typically the ones redesigning underlying processes alongside AI deployment rather than simply layering new technology onto inefficient workflows.
The Pivotal Role of Open Source
Open source is becoming increasingly important to ROI because it gives firms more interoperability and visibility into their software supply chain. This becomes particularly important as firms try to manage the cost and scarcity of specialized AI infrastructure. An enterprise open source platform can give firms the freedom to run any workload in any environment, at any cloud provider, over any silicon architecture (e.g., CPU, GPU or QPU), reducing dependence on expensive GPU resources and the risk of excess technical debt. It allows firms to run workloads wherever and whenever it best suits their needs. This provides greater control over how and where workloads are deployed.
Open source AI models can also improve transparency. In financial services, organizations need to understand how models were trained, what data was used, how decisions are made, and how outputs can be explained to regulators and customers. Proprietary models can create significant challenges when institutions need to test, govern, and possibly defend the fairness or accuracy of an AI-driven decision. Red Hat’s guidance on AI in banking highlights many of the same challenges around scaling AI in financial institutions, including data, compliance, operations, and governance.
Broader industry initiatives, such as the Open Accelerator project, are also helping to support more open and responsible AI development by encouraging cross-sector collaboration around open source AI, reusable tools, and practical frameworks for innovation. These efforts ensure AI develops in a way that is transparent, portable, and accessible across the broader ecosystem.
Conclusion
Financial services firms need clearer business objectives, stronger governance, and more disciplined deployment strategies to use generative and agentic AI at scale. The stakes are high in financial services because AI decisions can affect customer trust, financial well-being, regulatory compliance, and even market stability. Against this backdrop, the firms that succeed will not necessarily be the ones moving fastest. They will be the ones building AI capabilities that are transparent, resilient, scalable, and worthy of the critical role they play in the financial system.
The author, Monica Sasso, is the digital transformation lead, for global financial services at Red Hat.