Anyone who has tried to lodge a development application in New South Wales knows the drill: a thicket of planning controls, zoning maps, setback rules and acronyms that seem designed to keep ordinary residents at arm’s length. Into that gap has stepped DAISY, an artificial intelligence planning assistant that councils across the state are now rolling out after it quietly racked up more than 16,500 conversations with residents trying to make sense of the rules that govern their own backyards.
According to Tech Business News, the chatbot is being adopted by a growing number of NSW councils, where it answers questions about planning information and walks residents through the early steps of preparing a development application. The headline number, that 16,500-plus figure, is less interesting for its size than for what it reveals: a very large appetite among ratepayers for plain-English answers to planning questions, delivered at the hour that suits them rather than the hour the counter happens to be staffed.
Why councils are reaching for AI
Local government in Australia has long been the Cinderella of the public sector, carrying a heavy load of frontline services on comparatively thin budgets. Planning and development is one of the most contested parts of that load. Residents want certainty about what they can build, neighbours want a say, and councils are under constant pressure from the state government to clear development applications faster. The Minns government has made planning reform and housing supply a central political theme, and slow approvals are regularly blamed for choking the pipeline of new homes.
A tool like DAISY sits at the very front of that process, before an application is ever formally lodged. If a resident can learn early that their proposed granny flat falls foul of a height control, or that a particular property sits in a heritage conservation area, they can adjust the plan before spending money on drawings and fees. For council staff, every question answered by the assistant is one fewer phone call or counter visit tying up a planner who could be assessing applications instead. That is the efficiency logic driving adoption, and it is a familiar one across government: use automation to soak up the repetitive, high-volume enquiries so that scarce human expertise is reserved for the genuinely complex cases.
The appeal is sharpened by the fact that planning questions are, on the surface, well suited to AI. Much of the information a resident needs already exists in written form, scattered across local environmental plans, development control plans and state planning policies. Stitching those documents together into a clear answer is exactly the kind of retrieval-and-summarisation task that modern language models do reasonably well, provided they are pointed at the right source material and kept on a tight leash.
Two ways of reading the trend
Supporters of the rollout see it as a sensible, low-risk way for councils to modernise a service that has frustrated residents for decades. On this view, DAISY is not making planning decisions; it is helping people understand a system that was always meant to be public but was never especially accessible. If it lifts the quality of applications that eventually land on an assessor’s desk, the whole process speeds up and fewer applications bounce back for missing information.
There is a more cautious reading, and it deserves equal airtime. Planning advice carries real consequences. A resident who acts on a confident but incorrect answer from a chatbot could sink money into a design that is never going to be approved, or worse, build something that lands them in a compliance dispute. Large language models are known to produce plausible-sounding errors, and planning law is riddled with exceptions, discretionary judgments and site-specific quirks that resist tidy automation. The question of who is accountable when an AI assistant gets it wrong, the council, the vendor, or nobody, is not one the technology answers on its own.
Governance watchers will also want to know how these tools handle the boundary between general information and formal advice. Councils have historically been careful to frame counter conversations as guidance only, with the formal determination reserved for the assessment process. An AI assistant needs the same guardrails, clearly signalling its limits and steering residents to a human when the question moves beyond the straightforward. The 16,500 conversations are a useful proxy for demand, but they are not, on their own, evidence of accuracy.
What it means for Australia
NSW is the obvious bellwether here, but the implications run well beyond it. Every state and territory wrestles with the same tension between faster approvals and careful decision-making, and local councils everywhere are watching for proven ways to do more without ballooning headcounts. If DAISY demonstrates that a planning assistant can take pressure off frontline staff while improving the resident experience, councils in Victoria, Queensland, Western Australia and beyond will face pressure to follow suit.
The experiment also feeds into a broader national conversation about AI in the public sector. Governments at every level are moving from cautious pilots to actual deployment, and planning is a telling test case because it is high-volume, document-heavy and emotionally charged. Australia still lacks a single, settled rulebook for government use of AI, with agencies leaning on a patchwork of assurance frameworks and procurement guidance. Tools like DAISY will test whether those frameworks are fit for purpose when the AI is talking directly to citizens about decisions that affect their property and their finances.
There is a competitive dimension too. The market for govtech and planning software is maturing quickly in Australia, and a product that can show genuine traction across multiple councils has a strong story to tell both at home and in comparable markets overseas, where the challenge of accessible planning information is universal.
What’s next
The near-term question is simply how far and how fast the rollout goes, and whether the councils adopting DAISY publish anything meaningful about outcomes rather than activity. Conversation counts are easy to report; the harder and more valuable metrics are whether applications are arriving in better shape, whether assessment times are falling, and how often residents are escalated to a human because the assistant could not help. Expect scrutiny, too, on how the tool is kept current as planning controls change, since an assistant trained on yesterday’s rules is a liability, not an asset.
For now, DAISY is a reminder that some of the most consequential AI deployments in Australia are happening not in glossy tech launches but in the unglamorous machinery of local government, where millions of residents actually meet the state. If it works, it will quietly change how Australians interact with the planning system. If it stumbles, it will offer an equally instructive lesson in the limits of handing public-facing advice to a machine.
Sources: Tech Business News
