Case Study

How Agentic AI-enabled delivery accelerated CI/CD modernization and created 40% savings

UST combined its Agentic AI application know-how and industry-leading DevOps expertise to transform a complex enterprise migration into a scalable AI-powered delivery engine

Impact at a glance

OUR CLIENT

Modernizing software delivery across one of Europe's largest retail organizations

Our client is one of Europe's largest home improvement and DIY retail groups, operating multiple consumer and trade brands across the continent.

As a FTSE 100 business with approximately 70,000 employees worldwide, speed-to-market directly impacts competitiveness. The organization depends on digital products, engineering teams, and modern software delivery capabilities to support its omnichannel business model.

THE CHALLENGE

A fragmented CI/CD estate made modernization too expensive to execute manually

Years of rapid expansion produced a DevOps environment that had never been fully rationalized. Multiple Jenkins instances operated independently, each containing unique configurations, scripting patterns, plugins, and shared libraries.

At the same time, GitLab had already been introduced as a strategic platform investment. Instead of simplifying the environment, it created a dual-platform ecosystem that increased operational complexity.

The organization faced several challenges:

• Fragmented CI/CD environments increased operational overhead and duplicated effort
• Hundreds of customized pipelines created standardization challenges
• Parallel Jenkins and GitLab environments increased governance complexity
• Manual migration approaches were slow, expensive, and difficult to scale
• Strict change controls required deterministic, traceable migration paths
• Internal teams lacked the bandwidth to execute a program of this size

Without intervention, the organization risked prolonging technical debt while delaying the return on its GitLab investment.

The business needed a different delivery model, not simply a faster migration project.

THE TRANSFORMATION

Proving AI could analyze at scale, then using it to build a repeatable migration engine

UST reframed the initiative from a one-off migration project into an AI-enabled transformation program; one designed to get more efficient with every phase, not just faster in isolation.

Earning the mandate with a live proof point

The client's leadership was evaluating a range of AI-assisted approaches to tackle its CI/CD estate and had not yet committed to a delivery partner for the bulk of the work. UST PACE FlowCraft, UST's Agentic AI accelerator, was put to the test directly: within a matter of days, it analyzed more than 1,000 Jenkins pipelines across multiple instances, mining components, dependencies, and pattern similarities that would have taken a manual team weeks to surface.

That demonstration earned UST the mandate to take on a significantly larger share of the migration in the next phase, at comparable complexity but roughly five times the volume.

From manual baseline to compounding AI efficiency

The program unfolded in phases, and the contrast between them tells the real story:

• Phase one established the baseline. It covered one complex repository and was delivered using traditional, largely manual engineering methods, before UST's Agentic AI approach was introduced. It set the reference point for scope, effort, and cost that every later phase would be measured against.
• Phase two introduced the Agentic AI-driven approach at a meaningfully greater scale, several times the repository and pipeline volume of phase one. Even with that scope increase, delivery time and cost grew only modestly relative to phase one, a direct result of AI-accelerated analysis, conversion, and validation.
• Phase three expanded scope to the largest volume yet. Despite the further increase in repositories and pipelines, cost and timeline grew only nominally compared to the scope. The clearest evidence that the AI-powered delivery model, not just team effort, was driving the efficiency curve.

Across phases two and three, UST benchmarked actual AI-accelerated delivery against its own granular manual-delivery estimates for the same scope. That comparison is the basis for the effort and cost savings cited in this case study.

AI-powered analysis at scale

UST PACE FlowCraft helped analyze thousands of Jenkins pipelines to identify common patterns, reusable components, complexity clusters, and opportunities for standardization.

This reduced analysis effort by approximately 67% compared to manual discovery methods and established a reusable migration blueprint carried forward into later phases.

Componentization instead of one-for-one migration

Rather than recreating every pipeline individually, teams standardized reusable architectures. The program established common components, skeletal pipeline structures, and reusable GitLab templates that could be used by multiple engineering teams.

This approach delivered approximately 50% pipeline reusability while reducing future maintenance requirements.

AI-assisted delivery execution

Agentic AI accelerated engineering activities throughout execution. Teams used it to accelerate:

• Code conversion
• Pipeline transformation
• Validation activities
• Delivery acceleration
• Documentation generation

As migration volumes increased phase over phase, the benefits of the AI-powered approach compounded rather than diluted.

Governance embedded throughout delivery

UST maintained governance from discovery through production cutover. Every migration activity remained traceable, deterministic, and audit-ready without slowing delivery.

The result was not simply a successful migration. It was a scalable delivery model that became more efficient as volumes increased.

THE IMPACT

Faster delivery, lower effort, and an AI operating model that improves with scale

The engagement delivered measurable business outcomes across productivity, cost, and operational scalability.

Beyond the metrics, the organization fundamentally changed how modernization programs can be executed.

The GitLab investment now operates as a consolidated engineering foundation rather than a secondary platform running alongside legacy systems.

Teams can now deliver faster, govern more effectively, and scale future migrations without proportionally increasing costs.

Enterprise modernization programs often fail because complexity scales faster than engineering capacity. This engagement demonstrated a different model: prove AI's capability at scale first, then use it to build a repeatable delivery engine. By combining AI, engineering expertise, and governance, organizations can transform large-scale migrations into reusable operating capabilities.

As programs scale, AI compounds value rather than adding complexity.

Explore how UST PACE DevOps accelerates enterprise migrations.