The panel, hosted by Bidwells Partner, Planning, Jamie Sullivan, bringing together voices from government, architecture, planning practice, and the innovation community explored a simple but transformative question: can AI help us build a planning system that is not only faster, but fundamentally better?
What emerged was a picture of a sector on the cusp of profound change.
A System at Breaking Point Meets a Technology Hitting Its Stride
Jamie Sullivan opened the session with a candid assessment. Planning officers are handling hundreds, sometimes thousands, of applications a year. Local plans, intended to refresh every five years, routinely take a decade. Political churn plays its part, but the deeper issue is structural: the system is being asked to do more than it was ever designed for.
AI is arriving at exactly the moment the system needs it most. And crucially, the innovation is no longer confined to the private sector. Local authorities, central government, and public‑sector partners are now driving some of the most exciting experimentation.
From Chatbots to Autonomous Agents
The panel: with Russell Curtis of RCKa Architects, Milan Bogunovic of MHCLG and Krithika Ramesh of Connected Places Catapult, noted that we have moved beyond the era of “one prompt, one output.” AI agents can now perform multi‑step tasks, interpret complex geospatial data, and synthesise evidence at a scale and speed that would have been unimaginable even two years ago.
This shift is not incremental. It is foundational. It means planners can spend less time wrestling with documents and more time applying judgment, design thinking, and local knowledge - the things humans do best.
The Tools Already Changing the Landscape
Across the UK, a new generation of AI‑enabled tools is emerging. Some extract decades of planning data from PDFs and map it for easy interrogation. Others automate the validation of applications or model infrastructure scenarios. Google’s new augmented planning tool, currently being tested with local authorities, aims to pull together all the information a planner needs for a householder application and draft a report in minutes.
These are not prototypes. They are working tools, already in use, already saving time. One of the most striking examples came from architect Russell Curtis, whose team at RCKa used geospatial AI to analyse more than two million data points across London’s freeholds. The result was a comprehensive map of small‑site capacity across the capital, revealing the potential for around 850,000 homes.
This evidence base is now informing the next London Plan. It enables a rules‑based small sites design code, supports borough‑level intensification strategies, and opens the door to new delivery mechanisms such as Local Development Orders. It represents a shift from treating small sites as unpredictable “windfall” to planning for them strategically. It is a glimpse of what becomes possible when data replaces assumption.
AI as a Bridge Across Sectors
Krithika Ramesh of Connected Places Catapult highlighted another breakthrough: AI can now integrate land use, transport, energy, and environmental data into a single decision‑making model. For the first time, planners can explore how housing targets interact with net‑zero goals, or how transport accessibility shapes energy demand.
This kind of cross‑sector optimisation has always been desirable. It has simply never been feasible at scale. AI changes that.
Augmented, Not Automated, Decision‑Making
Milan Bogunovic from MHCLG emphasised that the goal is not to automate planning decisions but to augment them. The new tools being developed with partners like Google are designed to gather evidence, analyse patterns, and draft recommendations, but the planner remains firmly in control.
The aim is to free planners from the administrative burden that currently consumes so much of their time, allowing them to focus on design quality, community engagement, and strategic thinking.
The Rules‑Based Debate
A recurring question was whether AI inevitably pushes the system toward a more rules‑based approach. The panel’s view was nuanced. Some standardisation is essential, particularly around data formats. Without it, AI cannot operate safely or effectively. But local discretion remains vital. Leeds is not London, and planning must reflect that.
The real risk is not over‑standardisation but poor data. As Krithika noted, the UK is making hundreds of billions of pounds of infrastructure decisions using inconsistent datasets. Standardisation is not a threat to planning judgment, it is a prerequisite for good planning.
The Real Prize: Rethinking the System Itself
Perhaps the most compelling insight came from Milan: AI should not simply speed up existing processes. It should allow us to redesign them entirely. If consultation responses can be processed in minutes instead of days, if evidence bases can be generated in days instead of months, if site capacity can be mapped city‑wide in real time, then why would we keep the same workflows?
AI is not about doing the same things faster. It is about doing different things and building a planning system fit for the 21st century.
A Future Already Taking Shape
The conversation at the Bidwells Pavilion made one thing clear: AI is no longer a future aspiration for planning. It is already here, already delivering value, and already reshaping what is possible.
The question now is whether the sector will seize the opportunity to shape this transformation or be shaped by it. The next five years will define the next fifty.
