"We have critical legacy systems nobody fully understands."
That line rarely comes up in a status meeting. It comes up when a deal is on the table, a platform is being inherited, or a rebuild decision needs sign-off and nobody can say with confidence what the system actually does, what it would cost to replace, or what breaks if it's touched. At that point, the real problem isn't technical. It's that a business decision is being made without the information needed to make it safely.
This case study looks at that moment specifically: not how fast a legacy system can be rebuilt, but how an organisation got the clarity to decide whether and how to rebuild in the first place working with a technology consulting partner who paired AI-powered analysis with on-demand specialist capacity to turn an unknown risk into a scoped, priced plan.
+ The Challenge
An organisation in the life insurance sector held a legacy Quote & Apply system that nobody fully understood more than 100,000 lines of blended-stack code, accumulated over years, with no reliable documentation of what it did or why. The context sharpened the stakes: the assessment was needed as part of an acquisition-related evaluation involving a reinsurer, where the cost of misjudging the system's scope wasn't hypothetical it would shape deal terms. This wasn't just a technical debt problem.
Bringing that kind of legacy assessment capability in-house would have meant hiring specialists the organisation didn't need permanently, or waiting months for a traditional discovery phase, neither of which fit the decision timeline. This is a common reason organisations look to staff augmentation services or an outside IT solutions company rather than building the capability internally.
+ The Solution
Rather than treating "understand the system" and "rebuild the system" as one long project, the two were separated. Working with an IT partner offering staff augmentation on demand, the organisation brought in AI purpose-built for legacy assessment: an agent fleet read the source code directly and auto-generated documentation, test coverage, and a stack-agnostic model of how the system actually behaved - no interviews, no waiting on institutional memory that may not even exist anymore.
With that model in hand, humans not AI made the calls that matter: what to keep, what to retire, what to reshape, and what a resulting rebuild would realistically cost and take. The engagement delivered a working proof of concept, evaluating translation of the legacy system into the target stack often OutSystems or another low-code platform giving leadership a costed, evidence-based path forward before committing to a full rebuild. A full AI-accelerated rebuild remains available as a distinct next step once that decision is made, rather than something the team moves into automatically.
+ The Results
+ 75% reduction in legacy analysis and design effort: the discovery phase that typically stalls modernisation and acquisition decisions was compressed dramatically.
+ A priced, evidence-based rebuild decision: a working proof of concept evaluating translation into the target stack gave leadership a costed view of the modernisation path without committing to a full rebuild before that decision was ready to be made.
+ Acquisition and integration risk reduced: the unknowns that make legacy platforms hard to value or integrate were surfaced up front, before commitment rather than after.
+ Proven at scale in the public sector: in a related engagement, the same AI-powered assessment approach delivered an 85% reduction in expected effort to replace 600+ legacy forms, for a government agency with strict compliance and continuity requirements.
+ No permanent hire required: working with an on-demand IT partner meant the organisation accessed niche legacy assessment and OutSystems rebuild skills only for the length of the engagement, rather than building an in-house team.
