Industry
Banking
Banking Services Transformation: Unlocking Exponential Value
Highlights
- Embedding artificial intelligence (AI) into mainframe engineering helps BFSI institutions overcome fragmented legacy knowledge, analyse complex dependencies, and de-risk core change programmes.
- Deploying AI in mainframes alongside modern IDEs, Git-based source management, and DevOps pipelines unlocks measurable productivity and operational efficiency across the entire software delivery lifecycle.
- Adopting a structured, phased roadmap accelerates legacy modernisation, enabling banks and insurers to respond swiftly to market and regulatory shifts and accelerate innovation while building the resilience needed for high-volume financial services.
On this page
On this pageinpage
The case for AI in mainframe engineering
In the banking, financial services, and insurance (BFSI) industry, mainframe platforms remain central to routine operations because they combine resilience, security, and deep business logic accumulated over decades. As business demands evolve, BFSI firms spend significant time and effort on change-the-bank (CTB) initiatives involving enhancing the codebase. At the same time, BFSI firms are also under pressure to implement run-the-bank (RTB) change programs to maintain mainframe applications and fix issues to ensure smooth functioning and deliver consistent customer service.
To unlock AI’s potential to reduce friction in mainframe engineering, BFSI firms must embed it into the engineering model rather than treating it as a mere productivity tool.
Decades-old business rules of critical banking and insurance systems reside on legacy mainframe systems—while this is an advantage, it also causes friction, primarily driven by fragmented knowledge, complex dependencies, and varied delivery practices. Artificial intelligence (AI) coupled with modern integrated development environments (IDEs), enterprise-level unified version management, and open connectivity standards, has the potential to transform the way mainframe engineering teams develop, maintain, modernise, and support applications. To unlock AI’s potential to reduce friction in mainframe engineering, BFSI firms must embed it into the engineering model rather than treating it as a mere productivity tool.
The second of a three-part thought paper series on legacy modernisation in BFSI, this paper will examine how AI can be leveraged to improve mainframe engineering. The first paper explains the need and approach to building a comprehensive, AI-powered knowledge hub, while the third will dwell on exiting mainframes.
How does AI mitigate mainframe challenges in BFSI?
Let us examine the challenges that continue to constrain BFSI firms in successfully implementing CTB and RTB initiatives across mainframe systems; for example, in upgrading features for existing products or adding a new business capability. Table 1 depicts the challenges and how AI-enabled engineering can help mitigate them.
Table 1: Addressing core mainframe engineering challenges with AI
How does AI transform the mainframe engineering lifecycle?
AI has the ability to support all the principal phases of a mainframe engineering engagement (see Table 2). The specific capability mix will depend on the business priority, application criticality, control requirements and the maturity of the engineering platform.
Table 2: Driving modernisation and operational efficiency though AI across mainframe engineering phases
The AI capabilities highlighted in Table 2 can be applied to a complex application maintenance and development environment, where limited visibility into embedded business logic and application dependencies increases the effort and risk associated with change. Specific to the BFSI industry, rules governing a plethora of critical functions such as payments, lending, card processing, account servicing, underwriting, policy administration, and claims are often distributed across interconnected applications and legacy components. Even a seemingly minor change can affect multiple service lines, customer journeys, financial calculations, and regulatory processes. Incomplete impact analysis or incorrect implementation can result in service disruptions, transaction failures, delayed product launches, regulatory exceptions, and inconsistent customer outcomes. These issues ultimately increase operating costs, weaken service delivery, and negatively affect customer experience and trust.
However, applying AI to mainframe engineering is not just about leveraging AI tools but also making sure that the relevant foundations or force multipliers are in place to ensure seamless implementation. The key force multipliers are:
Embracing IDEs can help BFSI firms to respond quickly to changing customer expectations and regulatory mandates. By broadening engineering participation through a consistent and collaborative workspace, BFSI firms can accelerate product innovation and respond faster to regulatory change.
Moving from mainframe-based source code management to Git-based source code management at an enterprise level establishes traceability, transparency, collaboration, and control required to introduce business change with greater speed and confidence. It also creates a trusted foundation for DevOps automation and AI-enabled mainframe engineering at enterprise scale.
Mainframe connectors provide engineers with secure and reliable access from modern IDEs while preserving existing mainframe security, compliance, and governance controls. This in turn enables BFSI firms to extend modern engineering and AI-assisted workflows to mission-critical mainframe systems responsible for transaction processing, payments, credit, policy administration and other core functions.
Adopting the DevOps value stream provides banks and insurers with the controlled delivery foundation required to convert AI-assisted engineering outcomes into faster product launches, timely regulatory compliance, and more reliable customer service, in turn unlocking measurable business value.
Agentic workflows for mainframe engineering follow the same core principles as distributed application development, but must address the specialised technologies, dependencies, and controls of BFSI systems of record, such as core banking and insurance platforms, customer databases, and financial records.
All these components integrate to create a complete mainframe ecosystem play (see Figure 1).
Figure 1: Integrating AI components across the mainframe software delivery lifecycle
How can BFSI firms implement a roadmap for AI in mainframe engineering?
BFSI firms must establish a structured roadmap based on business priorities, application criticality, engineering maturity, and enterprise controls. Adopting a phased approach will empower mainframe engineering teams to leverage AI to the fullest extent possible. In addition, successful implementation of AI in engineering requires coordinated action from business leaders, architects, engineering teams, risk functions, technology vendors, and systems integrators. Let us examine the progressive phases of a AI-driven approach to mainframe engineering in more detail.
- Phase 1: Assess the current AI maturity to establish ambition and risk posture. Benchmark the AI maturity against a 5-stage maturity continuum.
- Phase 2: Empower engineering teams with modern development and delivery environments encompassing source code management (SCM), tools, and connectors.
- Phase 3: Automate, build, and deploy for mainframe application development.
- Phase 4: Build a mainframe knowledge hub comprising an engineering foundation to document the application business rules and technical artifacts.
- Phase 5: Run pilots to validate priority use cases, spanning impact analysis, testing, human oversight, and measurement of outcomes. The degree of AI usage will depend on AI maturity in mainframe engineering.
- Phase 6: Industrialise and institutionalise AI capabilities proven through pilots across applications and beyond to other use cases. Standardise agent patterns, establish uniform governance, and adopt enterprise-wide standards to move from pilots to reusable enterprise AI capabilities.
- Phase 7: Scale and optimise by expanding AI to other enterprise applications. Increase autonomy selectively, monitor drift and exceptions, and continuously improve knowledge, tools and policies.
This phased roadmap offers a practical path to begin the AI journey while steadily building towards a future-ready, AI-enabled mainframe engineering ecosystem.
Case-in-point
A large US-based insurer was facing challenges in understanding the business functionality embedded within complex mainframe systems and assessing the impact of change requests. The firm wanted to reduce the risk associated with introducing new changes. We developed a suite of custom, context-aware AI agents to streamline the development workstream by:
- Enabling developers to use natural language to interact with mainframe applications and receive contextual responses directly within their workspace.
- Reducing the impact analysis effort by 40-50% through better traceability.
- Accessing insights from previous programs and incorporating them into the solution approach.
- Improving developer productivity through an integrated platform comprising IDEs, DevOps infrastructure, and AI agents.
The future of AI-enabled mainframe engineering in BFSI
As BFSI firms continue their modernisation journeys, mainframes will remain an integral part of enterprise technology landscapes. Mainframe platforms must evolve alongside other emerging technologies. So, the future of mainframe engineering lies in integrating it with modern development practices and technologies.
The future of mainframe engineering lies in integrating it with modern development practices and technologies.
Going forward, the rapid advancement of autonomous software engineering agents coupled with the other AI technologies will reshape how existing mainframe applications are upgraded, analysed, remediated, tested, maintained, and modernised. The engineering teams of today will rely less on platform-specific expertise and more on AI-assisted workflows.
We believe that BFSI firms must prepare for this future by transforming their engineering practices, incorporating modern IDEs, SCMs, AI assistants, and DevOps pipelines, in addition to skills modernisation, and transforming the mainframe into an AI-augmented, developer-friendly platform that continues to power critical business functions. This will equip banks and insurers with the agility needed to respond quickly to market and regulatory changes, accelerate product innovation, and improve customer service. Most important, it will help infuse resilience into systems to support high-volume, time-sensitive financial services.