Client background
The client is a private equity-backed transportation company undergoing a large-scale digital modernization of the core ERP system and several mission-critical specialized applications.
Business challenge
The client aimed to accelerate development velocity, strengthen delivery timelines, and improve QA coverage across a multi-stream engineering program, all while managing unplanned scope increases and evolving requirements. At the start of the engagement, AI tools were active on fewer than one in seven engineers' desks, with no shared workflows, no adoption governance, and no measurement baseline in place. The organization sought to introduce AI tooling into its workflows to enhance the productivity of the 140+ engineering team without adding overhead. Achieving meaningful GenAI adoption at scale, alongside a step-change in test-driven development maturity, were the client's main goals.
Implementation
The delivery team helped the client design and execute a structured AI adoption program across the client’s six engineering work streams. The initiative was built to embed GenAI productivity directly into the software development lifecycle, rather than treating it as a standalone effort.
To achieve this, the delivery team helped the client with the following areas:
GenAI tool rollout at scale. The delivery team identified and deployed the most appropriate GenAI tools for each engineering role and use case, reaching all 140+ engineers across Java development, front-end, QA, data, and BI disciplines.
Use-case acceleration. The delivery team mapped and accelerated 39 distinct engineering use cases, achieving time improvements ranging from 1.01x to 144x depending on the task type.
SDLC integration. AI productivity was embedded primarily within coding and QA workflows, reducing manual coding effort, accelerating test creation, and cutting debugging and code review time.
Legacy codebase handling. The delivery team applied GenAI to reverse-engineering tasks on legacy business logic, reducing turnaround time from two weeks to two days, and to automated documentation generation, cutting it from approximately one day to around ten minutes. These were among the highest-impact single-use cases in the engagement and directly addressed the delivery risk posed by undocumented legacy systems.
Agile process integration. GenAI tooling was incorporated into existing Scrum and SAFe routines to ensure adoption was sustainable within the team's delivery cadence and did not create additional overhead.
Onboarding acceleration. The delivery team helped restructure new-engineer onboarding to leverage AI-assisted workflows, reducing ramp-up time from two weeks to three days.
Throughout the engagement, the delivery team addressed key adoption barriers, such as developer skepticism, implementing security guardrails in an enterprise environment, and creating an effective AI context for a complex legacy codebase.
Value delivered
GenAI tool adoption rose from 13% to 91% (a 78-point increase), being the largest adoption gap closed in a single engagement. This demonstrates that structured rollout and change management are as critical as the tooling itself;
Average team velocity grew from 85 to 108.3 story points (+27%), while delivery stability improved: milestone slippage, scope creep, and chronic overtime were all brought under control;
Test coverage increased from 55% to 81%, driven by a dramatic acceleration in test creation speed;
Code review time was halved from 8–12 hours to 4–6 hours, hotfix deployment time dropped from 4–6 hours to 1–2 hours (-70%), and survey data indicated a 30% reduction in manual coding time and a 50% reduction in debugging time;
Legacy reverse engineering was reduced from 2 weeks to 2 days (-85%), and documentation generation fell from approximately 1 day to around 10 minutes.
Expertise: Generative AI development and AI development services.
