DWP casework assistance tool wins civil service competition

A DWP casework assistance tool has won a GDS competition centred around building AI and data solutions to frontline challenges.

The Casework Compliance Assistant, put forward by DWP Compliance Officer Marlon Woodley, was placed first by the judging panel at the grand finale of the 2026 Civil Service AI & Data Challenge this week.

The tool, which seeks to reduce errors and accelerate decision-making when accessing benefits claims for discrepancies, inaccuracies, or potential non-compliance, was created in a few hours though “vibe-coding”. It is now guaranteed £50,000 of development work from programme partner NTT DATA, preparing it for implementation.

Ian Murray MP, the Minister for Data and Digital Government, announced the winner at Church House in Westminster. He said: “Marlon’s standout idea has real potential to transform how government serves people.

“By giving caseworkers better tools, it could help them make faster, more accurate decisions and focus their time where it matters most – improving our public services.”

The idea was favoured over the seven others selected to participate in the hackathon semi-final and final.

This selection took place following a workshop and submissions process, in which the feasibility and value of long-listed ideas were tested, and nascent teams invited to put forward combined submissions. Following this, judges equipped each of the eight successful teams with a variety of civil service volunteers to aid in its further development.

The Casework Compliance Assistant works by allowing compliance officers to input details of a potential act of non-compliance, which the system automatically classifies. The case then passes to the pre-interview stage, where the tool provides a summary of supplied and missing documentation, alongside allowing the caseworker to generate interview questions.

Once the process reaches the “Ready for Final Check” stage, the Casework Compliance Assistant can also generate a final summary of the case, which the caseworker is able to edit for accuracy.

The other ideas showcased in the final were an HMRC fraudulent document detection tool, a FOI request assistant submitted by a civil servant from the Department of Energy Security and Net Zero, and an application creating AI-generated personas to test how different audiences would respond to selected policies, from the Department of Energy Security and the FCDO.

Previous winners include a former prison officer who conceived of a platform to streamline cell allocation, a HMRC tax compliance officer who proposed the development of a modern slavery dashboard, and Natural England team members who came up with a machine learning system which aids efforts to restore peatland.

Featured  Image Credit: Government Digital Service