Swindon Borough Council was Highly Commended in the Artificial Intelligence category of the Housing Technology 2025 Awards. Arlene Griffin explains how the council used AI during the replacement of its housing management system.
At first, the idea of replacing our legacy housing management system seemed deceptively simple. All we had to do was choose the right system, migrate our existing data, integrate the two and that would be that.
Of course, we’re not quite as naïve as that – we knew it wouldn’t be quite so straightforward. We anticipated complexity. We understood the NEC system we’d selected was powerful and required heavy configuration. We even knew we’d need external implementation skills to support us. What we underestimated was the true scale of the task ahead.
Despite having NEC consultants and our implementation partner QLP onboard, a significant amount of the work still rested on the shoulders of our internal project team drawn from across the business.
On time & within budget
It soon became clear that to meet our delivery milestones and stay within budget without compromising on best practice, we would need to significantly reduce our manual efforts.
The main question was: how do we deliver ‘right first time’ while cutting 100s of people-hours from the project?
Exploring the potential of AI
QLP proposed a proof of concept (PoC) to see whether AI could streamline some of our most time-intensive processes, particularly the transformation of our business requirements into clearly defined, testable acceptance criteria. This had previously been an entirely manual task:
- User stories created by business analysts;
- Development of the acceptance criteria;
- Review and sign-off by business stakeholders;
- Test cases manually written by the test team.
Automating acceptance criteria
The PoC started by automating the creation of acceptance criteria for each user story. The early results weren’t great because the AI-generated outputs lacked accuracy, invented requirements and used incorrect terminology.
But iteration of the acceptance criteria paid off. With enriched prompts that included housing-specific legislation, user personas from front-line teams, best-practice examples, logical constraints and clear formatting instructions, the quality improved dramatically.
Where a business analyst might typically take eight minutes to draft a single set of acceptance criteria, our AI could now generate 12 accurate, comprehensive criteria in under a minute.
Quality over speed
The principle of ‘right first time’ isn’t just about saving time, it’s about getting it right from the start. Speed alone wasn’t enough; we needed assurance that the outputs from QLP’s AI met our real business needs.
Fortunately, they did; the new format, clearly structured and grouped by business process, enabled our people to conduct rapid reviews. Not only did the acceptance criteria meet our expectations but in some instances they also highlighted gaps in the process that our teams hadn’t previously considered.
In short, the ‘human in the loop’ review process raise the quality from unacceptable to exceptional.
Test-case automation
With generation of the acceptance criteria working smoothly, the next step was to apply the same AI-prompt structures and personas to create system-agnostic test cases.
These detailed, step-by-step tests could be quickly validated on the live system, allowing testers to add any system-specific instructions.
The impact? A task that previously took up to two hours per test case now took less than a minute.
From PoC to full adoption
The success of the PoC led to full adoption of the methodology. What began as an experiment is now a key part of our project delivery toolkit.
The efficiencies gained have been substantial. We’ve saved the equivalent of over two person-years of work so far:
- HMS acceptance criteria: from six minutes (manual) to less than one minute (AI), saving 30 days;
- HMS test cases: from 90 mins (manual) to less than one minute (AI), saving 780 days;
- Total person-days saved: approx. 810 days overall.
Arlene Griffin is the head of housing commissioning and strategy at Swindon Borough Council. The council was Highly Commended in the Artificial Intelligence category of the Housing Technology 2025 Awards.
