Modernising Legacy Energy Applications in the UK: From Black-Box Risk to Operational Resilience
For UK energy leaders, legacy modernisation is rarely a back-office technology exercise. It is a business continuity decision. Across generation, trading, plant operations and supporting engineering functions, many organisations still depend on applications built years—or decades—ago. These systems often continue to do important work, yet they are increasingly hard to maintain, difficult to explain and risky to change.
The problem becomes acute when the application is undocumented. Source code may be unavailable. The original developers may have moved on. Dependencies may be unclear. What remains is a system the business still relies on, but no longer fully understands. In the UK energy sector, where operational uptime, safety, resilience and governance all matter, that creates a very specific form of executive risk: the software still works until the day the business needs to change it.
This is where AI-assisted modernisation is proving its value—not as a black-box shortcut, but as a controlled way to recover, explain and modernise opaque applications faster.
Why undocumented legacy systems matter so much in UK energy
Energy companies in the UK operate in an environment shaped by constant pressure to improve resilience, manage costs and adapt infrastructure while maintaining reliable day-to-day operations. That makes aging applications more than a technical inconvenience. They can slow operational change, delay standardisation across sites and increase dependency on shrinking pools of specialist knowledge.
Many of the most exposed systems are not necessarily the largest platforms. Often, they are smaller operational tools used for maintenance, monitoring, engineering workflows or local plant activity. Because they are business-critical but poorly documented, they tend to remain untouched for years. Over time, that creates concentrated risk:
- updates become harder and slower
- security and compliance concerns become more difficult to address
- rollout to other sites becomes impractical
- resilience depends on systems no one can confidently explain
For executive teams, the question is no longer whether to modernise, but how to do so without introducing unnecessary disruption.
A practical proof point from RWE
RWE Generation Ltd faced exactly this challenge with a 24-year-old application called Tube Tracker, used to help manage pipe systems in power plants. The application was important to operations, but it had become a classic black-box dependency: written in Java, lacking accessible source code, missing documentation and without experts left to maintain it.
Instead of treating this as a long, high-risk rewrite, Publicis Sapient and RWE used an AI-assisted, human-controlled approach to recover and modernise the application in just two days.
The sequence was straightforward but powerful:
- **Decompilation** of binary files into readable Java source code using open-source AI tools.
- **Rebuild** of the application in a modern environment using Java 17 and PostgreSQL 16.
- **Refactoring** of the codebase to improve readability, consistency and maintainability.
- **Business logic extraction** to reveal the application’s entities, data flows and core functionality.
- **Documentation generation** so future teams could understand and extend the system.
Throughout the work, AI accelerated the heavy lifting, but engineers remained in control. Outputs were reviewed, validated and shaped by human judgment at each critical step.
That distinction matters. In energy, leaders do not simply need speed. They need confidence in what was recovered, what was changed and what can now be governed.
What executives should take from this
The Tube Tracker modernisation was not just a successful application rescue. It demonstrated a broader operating model for UK enterprises dealing with fragile legacy estates.
First, it showed that undocumented systems do not always require a costly, speculative rewrite. With the right approach, organisations can recover technical understanding quickly and turn an opaque dependency into a maintainable asset.
Second, it proved that AI is most valuable when paired with engineering oversight. Black-box automation may create anxiety in regulated or operationally sensitive environments. Human-in-the-loop delivery changes the equation by making the process transparent, reviewable and governable.
Third, it highlighted the business value of modernisation beyond code conversion. In RWE’s case, the application became deployable, maintainable and suitable for rollout across additional sites. That means modernisation supported not only technical renewal, but also operational continuity and future scalability.
The measurable outcomes were significant:
- one engineer completed the work in two days rather than roughly two weeks of manual effort
- automated code generation delivered time savings of 35% to 45%
- test creation and unit test setup improved by 30% to 40%
- the codebase was reduced from roughly 7,000 lines to 5,000 lines through cleaner modern syntax
From one rescue to a repeatable UK modernisation model
Most UK energy organisations are not managing a single brittle application. They are managing portfolios of them. That is why the strategic opportunity is bigger than rescuing one system at a time.
A more scalable model starts with triage. Leaders should identify applications where five conditions overlap:
- high operational importance
- low maintainability
- obvious technology obsolescence
- active business pressure for change
- disproportionate continuity or governance risk if failure occurs
Those are the applications most likely to benefit from AI-assisted recovery and modernisation.
From there, the goal should be to build a repeatable pipeline: recover the code, explain the logic, generate documentation, rebuild on a modern stack, add tests and prepare the application for safe deployment. Done consistently, this creates more than project delivery. It creates a modernisation capability.
Why operating model matters as much as technology
One of the clearest lessons from Publicis Sapient’s work is that tooling alone is not enough. Faster outputs only become valuable when teams can trust them, review them and adopt them within a delivery model built for transparency.
That is why integrated teams, agile ways of working and visible governance are so important. Product, engineering and business stakeholders need to work from the same understanding of what the application does today, what must be preserved and what should change. AI can compress effort across analysis, code transformation, testing and documentation, but trust comes from the operating model around it.
For UK energy leaders, this is the real prize: not simply quicker migration, but a more resilient way to modernise critical systems without losing control.
Turning legacy risk into a maintainable asset
The most dangerous applications in an energy estate are often the ones that seem too awkward to touch. They remain in place because no one wants to disturb them, even as risk accumulates. Yet these are often the very systems where AI-assisted modernisation can create the fastest and most visible value.
RWE’s experience shows what is now possible. A 24-year-old black-box application with no accessible source code and no documentation was transformed into a readable, maintainable and deployable asset in days, not months.
For UK executives facing similar challenges, the message is clear: legacy applications do not have to remain opaque liabilities. With AI accelerating the work and humans firmly in control, they can become understandable systems, stronger operational assets and a practical foundation for broader transformation.