Productivity is the Answer!

We all know that whatever the question, the answer is "£700million and 1,300 extra staff" but actually the real answer appears to be 'automation and productivity'. This from an NAO report published last week:-

Productivity improvement: The Justice Transcribe AI‑enabled transcription tool

This case study focuses on one example of a productivity improvement to automate and improve processes, where the Ministry of Justice (MoJ) and HM Prison and Probation Service (HMPPS) sought to improve productivity within the Probation Service by introducing a transcription tool, Justice Transcribe, for probation officers to use in meetings. MoJ and HMPPS expect that this tool will lead to productivity improvements by reducing the amount of time needed to write up meeting notes; improving the quality of those notes; and improving the quality of the interactions probation officers have in the meetings, because they are not focused on note writing. By reducing the time probation officers spend writing up supervision meetings they have with people on probation, Justice Transcribe makes more time available for activities that directly contribute to public protection and rehabilitation. The case study sets out how the tool was identified, developed, tested, scaled and embedded into operational practice by MoJ and HMPPS. It also highlights key lessons from each stage of implementation. 

The case study is intended as a practical example for civil servants and public sector leaders considering similar productivity initiatives, providing insights into the decisions, governance arrangements, testing approaches, implementation activities and evaluation methods that supported delivery. This publication is the first of a series of case studies we plan to publish highlighting practical steps public bodies have taken to improve their productivity and the lessons others can learn from them.

The capacity challenge in the Probation Service 

1 The Probation Service has been under significant strain from staffing shortfalls and increasing demand. In our report Building an effective and resilient Probation Service, we found that, in July 2025, in 10 out of 12 probation regions, Probation Officers were working at above 100% capacity on average. Data from April 2024 indicated that the service had been operating with only around half of the staff needed to run sentence management as intended. In January 2026, the Sentencing Act was passed which introduced measures intended to increase the use of community-based sentences, placing further demands on the service. At the end of June 2026, only 74% of Probation Officer grade posts were filled. 

2 In February 2025 HMPPS established its ‘Our Future Probation Service’ (OFPS) programme. Its purpose is to improve performance, increase retention and reduce staff sickness absence. The programme aims to free up operational capacity by 25% by April 2027. Justice Transcribe, a transcription and summarisation tool enabled by artificial intelligence (AI), has been developed to reduce the amount of time probation officers spend writing up notes of the meetings they have with people on probation, improve the quality of the meeting notes and improve engagement in meetings. This case study sets out how MoJ and HMPPS developed and rolled out Justice Transcribe, and some key lessons for other organisations interested in a similar tool.

Identifying opportunities for productivity improvement 

3 Probation officers conduct approximately 3.7 million supervision meetings annually.3 An accurate record of the key points of these meetings is important for managing risks to the public and tracking progress with sentencing requirements. Writing up meeting notes and recording them in the case management system was time‑consuming and created a risk of poor quality and inaccurate records. 

4 The idea for Justice Transcribe came from front-line probation teams, who had identified the potential for AI-enabled transcription and summarisation to help them manage their workload and were exploring commercial tools on the market. HMPPS assessed the idea alongside wider organisational priorities and considered several factors, including the following.

Productivity improvement 

15 HMPPS expects productivity benefits from Justice Transcribe to include both time savings and quality improvements. It forecasts that Justice Transcribe will save around 450,000 hours per year for front-line probation officers. This estimate is based on an assumed time saving of 10 minutes per meeting on average and that the tool is used for 70% of the 3.7 million meetings a year. HMPPS also expects additional savings for support and other staff. As at April 2026, HMPPS estimated it had a capacity gap of 3,500 FTE probation staff. It expects Justice Transcribe will release time savings equivalent to 470 FTE probation officers, support and other staff, helping it to reduce this capacity gap. These estimates are based on modelling assumptions, and HMPPS told us it is planning further evaluation work to validate time savings in practice and collect more robust evidence of how the tool has changed the delivery of face-to-face supervision. 

16 HMPPS does not have actual time saved data to monitor how far estimated time savings have been achieved. However, it does have data on the number of meetings for which the tool has been used. Between 7 October 2025 and 14 September 2026, over 1,600,000 meetings were summarised using Justice Transcribe. HMPPS estimates that using Justice Transcribe has saved probation officers around 266,667 hours over this period.

17 HMPPS also holds feedback data from staff. Between launch and 2 June 2026, staff using the tool reported high satisfaction rates, with an average of 4.7 stars out of 5 across 63,000 in-app reviews. Probation officers we spoke to told us that Justice Transcribe saved significant amounts of time and improved the quality of meeting summaries as well as interactions in supervision meetings: 

“What it has allowed us to do is... have more meaningful conversations with the people we’re managing… being more active listeners, … rather than having our attention diverted elsewhere.” (Probation Officer) 

HMPPS also holds data on staff retention and sickness absence, so it could use this information to track whether uptake of Justice Transcribe is associated with benefits such as lower sickness absence or improved staff retention.

Further evaluation 

18 MoJ and HMPPS started an internal evaluation of the national rollout of the tool in 2026. This will include evaluation of how the tool is used in practice, how errors are identified and corrected, and an assessment of the quality of the final records. This national evaluation is an opportunity to build on the insights from the pilot and determine the full impacts of the tool, including the extent of time savings and quality improvements. It is also an opportunity to assess evidence of the benefits including, for example, improved retention rates or reductions in sickness absence.

Quality assurance of case management records

20 The pilot evaluation identified that there was a risk that meeting summaries were not adequately reviewed by probation officers. It recommended regular quality assurance audits and training on how to undertake a meaningful human review. Training and guidance materials emphasise the importance of reviewing the AI notes and the app included reminders for staff to do this at the point of use. For individual meetings, probation officers are expected to review the outputs from the tool before they add them to the national case management system.

21 MoJ and HMPPS are developing their quality assurance procedures. The Artificial Intelligence Playbook for the UK Government recommends that public bodies establish performance metrics for AI systems that measure the accuracy of model outputs.The national evaluation will include a linked case audit which will compare a sample of transcripts, AI-generated summaries and final case records to establish whether practitioners identify and correct errors or omissions and add relevant context and professional judgement. HMPPS also plans for regional Performance, Assurance and Risk teams to carry out case audits to look at the quality of the case records, alongside ongoing supervision by Senior Probation Officers that will include review of case management records. At present, however, there is no quality assurance mechanism to monitor the accuracy of the original transcript. This makes it difficult for MoJ and HMPPS to monitor the technical quality of the tool or the risk of probation officers becoming over-reliant on AI-generated summaries and not checking them. This could be addressed, for example, through periodic reviews of a sample of audio recordings, transcripts, summaries and final case records.