Background
Worcestershire County Council (WCC) has worked with OutSystems since 2015, when pressure on local authority budgets and the rising cost of providing social care prompted the council to look for new ways to deliver services more efficiently.
Digital channels became an important part of that response, allowing the council to reduce operational costs while meeting growing expectations for online public services.
Over the subsequent 11 years, WCC’s Digital Delivery Team has used the OutSystems platform to develop more than 100 applications, primarily public-facing services alongside applications supporting internal council processes.
More recently, the council has begun exploring how AI can extend that approach by reducing repetitive administrative work and making existing processes more efficient.
“AI was introduced to safely reduce repetitive administrative work and make processes more efficient,” said Jo Hilditch, head of digital, data and web services at Worcestershire County Council.
Challenge
One area WCC identified was the amount of staff time being spent on administrative processes such as manually verifying documents, checking whether information met the necessary criteria and contacting residents to correct errors.
Rather than deploying AI as a standalone technology, the council wanted to introduce it within clearly defined services and workflows where it could address specific operational problems.
Hilditch said the council’s aim was not to replace existing service design or staff decision-making.
“The goal was never to replace good service design or human judgment with AI. It was to make work faster, safer, and smarter.”
Governance was another consideration. WCC handles sensitive information and must comply with data residency requirements, influencing which AI models it can use. According to the council, strict UK-only data processing requirements have meant it cannot use some newer models, including GPT-5.
The council has instead embedded AI within defined applications with a clear business purpose, controlled data sources and known users. It can select different models according to individual use cases and monitor token allocation and usage limits to control costs.
Solution
One of the council’s first applications of agentic AI is its Bus Pass AI Agent, built on the OutSystems agentic systems platform.
The agent supports the concessionary travel pass application process by validating evidence submitted by residents before it reaches a human reviewer.
Previously, incorrect evidence could result in repeated contact between residents and customer service teams. The agent can identify problems with submitted documents at the point of contact, including photographs that do not meet the required criteria, incorrect proof-of-address documents and invalid dates of birth.
Human review remains part of the process, with the agent validating evidence before it reaches council staff.
In its first eight months, the Bus Pass AI Agent processed 2,498 applications and automatically validated 7,494 uploaded documents. It identified and rejected 643 invalid photographs, 364 incorrect proof-of-address documents and 286 invalid dates of birth.
If you liked this content…
By identifying those problems when information is submitted, WCC says it can prevent incorrect data from reaching human reviewers and reduce the administrative back-and-forth required to resolve applications.
Results
The AI work builds on wider changes WCC has made to how it develops and delivers digital services.
The council says using OutSystems has enabled its Digital Delivery Team to develop applications three to four times faster than through traditional software development, while supporting closer collaboration with business teams and reducing application maintenance costs. Some applications designed to support the local community have been delivered in less than 24 hours.
WCC has also shifted 80 percent of citizen contacts to self-service, which it says has saved hundreds of hours of manual processing.
AI is now being applied to further administrative processes. Alongside the bus pass agent, the council has developed a School Attendance Fines Validator, which it says saves up to 200 hours each year on document checking.
As WCC develops more AI-enabled applications on the OutSystems platform, the Digital Delivery Team is also building a library of reusable AI components and patterns that can be applied to subsequent services.
“The more that is developed on the platform, the more reusable AI components and patterns the team has, the faster the team is,” said Hilditch.
Governance and scaling AI
WCC’s approach is to keep AI within governed applications rather than allowing disconnected AI tools to proliferate across the organisation.
The OutSystems platform provides a centralised governance layer through which the council can maintain visibility of agents and embed AI capabilities into applications already used by staff.
For WCC, this also allows individual AI use cases to be assessed separately. The council can choose models according to the needs of a service – including lower-cost models where they are sufficient – while monitoring usage and expenditure.
Hilditch said councils considering a similar approach should begin with the service they want to improve rather than the technology itself.
“Any council considering AI should start with the service challenge, not the technology. They must be clear about the outcomes they’re trying to achieve, get their governance and data foundations in place, and focus on a small number of high value use cases that can demonstrate real benefits.”
Next steps
WCC sees AI as an extension of its existing digital transformation rather than a separate programme.
Its development approach increasingly combines low-code application development with reusable AI capabilities, allowing the Digital Delivery Team to introduce AI into established applications and workflows while retaining central oversight.
For other councils, Hilditch said the important lesson is to avoid treating AI deployment as an end in itself.
“The biggest mistake is treating AI as a technology project rather than a service transformation opportunity. Start small, learn quickly, and scale what works.”
