If you can incorporate AI into the complex and bureaucratic local government planning sector, you can do it for anything, says Brett Leahy at the London Borough of Redbridge. But where else can councils benefit? Quadrant Smart sits down with Leahy, head of building planning and control, to find out more
Tell us a little about your role in the council, and why you’ve looked to utilise AI project in your department?
BL: I think one of the main drivers in relation to starting to look at AI is in the context that our resources are diminished, and the demand has ever-increased. That’s true for local government, but it’s certainly true for planning and the expectations around planning to facilitate economic growth, social wellbeing, and to protect the environment.

We live in an age where you’ve got more planners that work in the private sector than work in the public sector. We also live in an age whereby the citizens’ expectations and their experience is more critical, particularly in having ready access to information, to contacts.
Back in 2008, 60% of our budget, our money came from central government, through the local government grant. Next year, it’s zero. We’ve had a 60% reduction in money that we have to spend on services. And I think the consequence of that is that it’s forced the issue for councils to cut back, but they’ve had to look at technologies as a means to still providing the citizen with the services they expect.
How is the council using AI for planning specifically in a bid to improve efficiency in the planning process?
When you submit a planning application as a homeowner, it’s checked by a technical team. They check the form, the drawings, various documents associated with that, like an Environmental Agency-type statement in terms of flood risk, or whatever it may be in terms of the property you live at.
The consequence of budget cuts is that it’s forced the issue for councils to cut back, but they’ve had to look at technologies as a means to still providing the citizen with the services they expect
What we’re trying to get the AI to do is actually check the drawings for us, check the documents for us, and then input that data on the database for reading application forms. We’ve been feeding the AI drawings and documents, getting it to learn. My team has been supporting Agile Applications, who provides the software – Agile gets the AI to do assessments and then they’re checked to see if it’s correct.
I suppose the AI is just quicker at actually reading information. It understands the criteria that’s acceptable. Because it’s all done in a live space: there is no manual ‘press this button, move this over to this bit, and press that button’. It’s all automated.
As the AI grows and learns, it’s doing it much quicker. We’ve indicated to save around 161 days per technical officer, per year. It’s a lot of saved time!
There are clear benefits in planning validation, but where else has the department incorporated AI for residents?
We have done a chatbot for residents as well. We get lots of emails and lots of queries, where people want informational guidance. Local government is bureaucratic, just by its nature. It’s still built on Victorian models of different hierarchies, in the way that they’re organised.
What we’ve found is, rather than asking us, they plug in the question for the relevant customer, and come back to us on it
As a user and citizen, you trying to get information, or just trying to understand what the process is, could find yourself pushed from pillar to post between various departments. We wanted basically to have the chatbot to help people navigate through a very complex planning system.
Previously, we would get hundreds of emails per day. We deal with 150 applications per week. We’ve got 15 staff dealing with around 150 a week. That’s loads of planning applications. The idea was that the chatbot will answer all of your planning questions in theory. It’s always learning. Any questions it doesn’t answer, we monitor it, and we then learn to provide that response, so its knowledge is ever-growing.

The bot has gone very well. what’s been useful about the bot is we have a call centre. Rather than a call centre passing on calls, they’re able to answer calls by using the bot themselves. What we’ve found is, rather than asking us, they plug in the question for the relevant customer, and come back to us on it.
What does the future of AI look for the future of local government, Brett?
Planning is one of the most complicated services around – if you can incorporate AI for that, you can do it for anything. Why don’t we do it for validation, which is the easier part of the planning process?
But then we want it to offer customers updates like you get from Amazon. If you took my building control team, I want you to have the Amazon experience. You know exactly where your building control application is at, and you get updates about what they’re doing about it.
The next stage of the planning validation project is then to do automatic consultation. When you as a homeowner submit your application, we’re required to consult certain individuals.
At the moment we have to interrogate what we call geographical information systems, which is basically a data-set or plan that shows different areas constrained, like flood, conservation, or an area of scientific interest. The next stage of the project is then for it to do automatic consultation, rather than someone manually doing that.
The final stage is basically to start to pre-populate various reports that we have to write. Say your neighbour wanted to submit an application and you wanted to object to it, you submit your comments – someone has to download that, and someone has to type those comments in. Why can’t we just get the AI to automatically do that for us?