Local plan consultations can generate large volumes of feedback from residents, developers, statutory consultees and community groups. For planning teams, reviewing, grouping and summarising these responses is essential work, but it can take significant officer time before the deeper analysis can begin.
Through the PropTech Innovation Fund, MHCLG’s Digital Planning Programme has supported a pilot exploring how artificial intelligence can help with this process, freeing up officers to focus on meaningful judgement, analysis and engagement.
The initial pilotGreater Cambridge Shared Planning (GCSP) and University of Liverpool looked to tackle this problem developing PlanAI, an AI tool designed to help planning teams summarise consultation responses. The results were promising. In the initial trial, the time taken to analyse planning consultation responses was reduced from around 18.5 hours to approximately 16 minutes.
As we set out in our previous blog post, the team were keen to find out whether this success could be replicated and scaled elsewhere in the country.
Scaling into other geographiesFive planning authorities from across the country took part in extending the pilot work to test whether the tool could deliver similar benefits in different settings:
Royal Borough of Greenwich Hull City Council Milton Keynes City Council Liverpool City Region Combined Authority Tameside Metropolitan Borough CouncilAcross participating planning authorities, the pilot produced efficiency gains of around 90%.
What did the councils learn?Many of the planning authorities agreed the tool helped them to summarise consultation responses. There were also useful reflections on how to get the most from the process and considerations on what they would do differently next time:
Involve governance teams early - Data security and information governance colleagues are critical and should be involved early in the process. In GCSP the information governance team recommended offering respondents an opt-out from AI summarising. In their most recent Local Plan consultation around 5% of respondents took that option, with those responses analysed manually in the traditional way. Planning teams felt this approach struck the right balance – giving people choice to avoid discouraging people from engaging in the plan making process. Think about structure before you start - If you’re going to apply a digital tool, you need to work backwards from the output you’re looking to get and reverse engineer your approach. For example, if you want the AI tool to accurately tag comments against specific policies, you need to make sure you have clear policy names and a well-structured document, so that the content can be easily coded and the tool behaves as expected. The team from the London Borough of Greenwich found that some tagging of comments to policies was inaccurate, and that overlapping policies created additional challenges. Getting your data architecture right before you begin will save significant time later. Manage expectations about timescales - The headline time saving figures are impressive, but it is important to remember this reflects one specific part of a wider process. Redacting responses, formatting data, and potentially going through several iterations to get useable outputs took two to three weeks for some pilot authorities. It is important for council teams to be upfront with their stakeholders this will still take time, especially in the early stages. Build in quality assurance - Think of the AI as producing a very good first draft - not a finished product. Planning officers still need to check outputs, and can go through several iterations to get the level of detail they need. Some thematic reports, for example, initially didn't distinguish between the volume of responses on different issues, which required adjustments to the model's parameters. There are efficiency gains from using the tool but quality assurance needs to be planned as part of the workflow. Be clear about what the tool does and doesn't do - The tool produces an impartial account of all responses. It doesn’t offer advice, recommendations, or weighting of views. While the project team see this as a strength, it needs to be clearly communicated to stakeholders and decision-makers so that everyone understands what they're looking at.Catherine McRory, Principal Planning Officer at The Royal Borough of Greenwich, said:
“We're on a very tight programme with our Local Plan with a three-month window to process consultation responses and prepare a revised plan. We received 3,000 comments from 600 respondents, so we were keen to explore where PlanAI could speed up that process while maintaining robustness. PlanAI proved to be a powerful support tool when dealing with large volumes of responses. But it's not a replacement for officer expertise and works best when combined with quality assurance and clear processes.”
The pilot proved that when it is used well, the tool can play a useful role in consultation analysis, particularly where councils are managing large volumes of consultation feedback within tight deadlines. However, the participating authorities learnt that the tool delivers the greatest value when combined with clear governance, well-structured data and professional oversight.
You can also watch this show and tell from February 2026 to hear how the pilots first got started.
Supporting innovation and choiceThis work reflects the wider approach of MHCLG’s Digital Planning Programme: testing promising digital tools with councils, gathering evidence about what works, and supporting the conditions needed for wider adoption. Alongside pilots such as this, the Digital Planning Programme is also developing a variety of AI tools for local planning authorities (LPAs). This includes Extract, an AI-powered tool that helps LPAs convert historic planning documents into usable, standardised data, and a new AI planning prototype is being tested with LPAs to support the processing of routine planning applications.
Together, these projects show how AI can be used carefully and practically in planning: not to make decisions in place of officers, but to reduce manual processing, improve access to information and help planning teams focus their expertise where it matters most.
The Digital Planning Programme is combining the development of shared national tools with funding and support for locally-led innovation. This provides both a common foundation and a diverse marketplace of solutions that LPAs can adopt to meet their own needs, working with industry partners to co-design along the way.
What happens nextThe next step is to continue building the evidence base for where AI can add value in planning, and identifying where additional safeguards, guidance or support are needed. As more planning authorities adopt and test tools such as PlanAI, the programme will continue to share learning openly so that councils can make informed decisions about the technology they use.
Through the Open Digital Planning network, MHCLG provides a space where LPAs can connect with peers who have already tested new approaches in their planning services, sharing practical guidance and evidence on return on investment to reduce the barriers to trialling and adopting digital tools.
The experiences of planning authorities in this pilot shows that, when used responsibly, AI can help them process information faster, strengthen evidence-led plan-making and give officers more time to focus on the decisions and conversations that shape places.
For more information about the Digital Planning programme, follow us on LinkedIn to stay connected, and subscribe to our newsletter for the latest updates.
seen at 11:33, 30 July in MHCLG Digital.