Digital Transformation of a Regional Bank

A regional bank had 800,000 customers, 20 years of transaction history, and a core banking system running on an IBM mainframe that the vendor had stopped supporting. Competitors were launching digital-first products. Regulators were pushing for improved digital channel access. And the internal IT team of 45 had built their careers on mainframe skills with little cloud exposure. The bank needed to modernise — but couldn’t afford to get it wrong.

The outcome: a working API layer over the mainframe within nine months, a mobile app launched on top of it, and 120,000 customers onboarded digitally in the first quarter after launch — without a single day of system downtime.

The Situation

The bank’s core banking system had been the backbone of operations since 2003. It processed two million transactions daily, maintained the ledger for every account, and calculated interest, fees, and regulatory reporting. It worked reliably — but it was a black box to everything outside it.

Every integration with the core system was a batch file process: overnight jobs that extracted data, transformed it, and loaded it into downstream systems. There was no real-time API layer. When customers called the contact centre, the agent was looking at data that was up to 24 hours stale. When the bank wanted to launch a mobile app, they had no way to query account balances in real time without replacing the entire core.

Three previous digital initiatives had started strong and stalled at the same point: they couldn’t get reliable, real-time data from the core system. The teams building digital products were blocked by the mainframe team’s queue.

The Challenge

The bank faced a structural problem that many traditional financial institutions recognise: the systems that worked well enough to build the business on were now actively preventing the business from evolving.

The specific constraints Nematix had to work within:

• Zero downtime tolerance — the mainframe processed two million transactions daily and could not be taken offline for migration
• No API surface — the core system communicated via COBOL batch jobs and CICS transactions; there was no REST or SOAP interface to build against
• Skills gap — the internal team had deep mainframe expertise but limited cloud, API design, or modern DevOps experience; any solution had to be operable by the team that would inherit it
• IBM contract expiry — the existing vendor contract expired in 24 months, creating a hard deadline on the longer-term migration plan
• Regulatory compliance — Bank Negara guidelines on data residency, audit trails, and digital KYC had to be met by any new digital channel

Our Approach

Nematix began with an eight-week discovery: mapping all 120 distinct banking processes, classifying each by frequency, complexity, and how tightly coupled it was to the mainframe’s internal logic. The output was a transformation heat map — which processes could be migrated quickly, which required the mainframe to remain the system of record indefinitely, and which could be solved with a facade layer.

The strategy that emerged had three phases.

Phase 1 — API facade layer (months 1–9)

Rather than replacing the mainframe — which would have taken years and carried enormous risk — we built an API gateway that translated modern REST requests into CICS transaction calls against the existing system. This gave the digital product teams an API to build against immediately, without waiting for a core migration.

The facade handled account enquiry, transaction history, balance lookup, and standing instruction management. It included a caching layer for non-real-time queries and a circuit breaker to handle mainframe queue saturation gracefully.

Phase 2 — Digital services on cloud (months 6–14)

With the API layer live, we migrated digital-specific functions — customer onboarding, digital KYC, document management, and notification services — to cloud-native services on Azure. These had no dependency on the mainframe’s data model and could be built cleanly. A new mobile app was built on top of the API gateway and these cloud services.

Digital KYC reduced the account opening process from a five-day branch visit to a same-day digital journey.

Phase 3 — Core migration planning (months 12–24)

In parallel with Phase 2 delivery, Nematix developed the roadmap for eventual core migration: evaluating modern core banking platforms, designing the target-state architecture, and building the business case for the board ahead of the IBM contract expiry.

Throughout all phases, we ran a structured upskilling programme alongside the internal IT team — pair programming, architecture workshops, and cloud certifications — so the bank would not be dependent on external expertise to operate what we built.

Outcome

The mobile app launched on time, built entirely on the API facade. The branch visit requirement for new accounts was eliminated for customers choosing the digital channel. The bank’s digital product team — previously blocked at every turn by the mainframe queue — now shipped independently.

Key Takeaways

Replace the interface before the system. The bank’s instinct was to replace the mainframe. The right first move was to put a modern interface in front of it. The facade pattern de-risked the project entirely and delivered value within months rather than years.

Upskilling is not optional. A modernisation programme that leaves the internal team unable to operate the result has failed, even if the technology works. Embedding knowledge transfer into every phase was what made the handover sustainable.

Batch is the enemy of digital. The overnight batch cycle was the root constraint — not the mainframe itself. Solving for real-time data access solved most of the digital product team’s problems without touching the core system at all.

This engagement draws on our Strategy & Transformation services. If your organisation is navigating a legacy modernisation challenge, start with a conversation.