AI in Development Application Processing: What NSW Councils Need to Know

Artificial intelligence is changing how councils process development applications. From document classification to privacy protection, AI can ease the load on planning departments. This article looks at practical uses for NSW councils, and at the governance that should come with them.

Several of these uses are available in the Local Government Hub, including AI Redaction.

The current state of AI in planning

Interest in AI across Australian local government is growing, and councils are looking at how it can ease workload pressure. The best results come from applying it to narrow, repeatable tasks.

Where AI adds value today

Current uses in planning focus on well-defined tasks:

• Document classification: categorising submitted documents
• Data extraction: pulling structured information from unstructured documents
• Privacy redaction: identifying and removing personal information
• Compliance screening: checking submissions against requirements
• Communication: drafting status updates and correspondence

These tasks share three traits. They rely on pattern recognition, they involve large volumes of similar items, and they support human judgment instead of replacing it.

Document classification and processing

Automated document classification is one of the most mature uses of AI for councils.

The challenge

A typical DA submission includes dozens of documents:

• Site plans and architectural drawings
• Survey reports
• Environmental assessments
• Statement of environmental effects
• Neighbour notifications
• Owner's consent details
• Supporting correspondence

Staff spend considerable time sorting, naming and filing these documents before assessment can begin.

AI-powered classification

Modern AI systems can:

• Identify document types from content, not just file names
• Extract key metadata such as addresses, dates and reference numbers
• Flag missing documents by comparing a submission with the requirements
• Detect duplicates and version conflicts
• Route documents to the right workflow stage

Accuracy varies by document type and by how consistent your submissions are, so measure it on your own documents.

Implementation considerations

• Start with high-volume document types: focus early effort on frequently submitted documents
• Plan for exceptions: set a clear process for documents the AI cannot classify
• Keep human oversight: staff should verify classifications, especially at first
• Track accuracy: monitor performance and retrain as needed

Automated redaction

Privacy protection is critical when publishing DA documents. AI-powered redaction speeds up this work while keeping a person in charge of the result.

Privacy requirements

Councils must protect personal information in publicly accessible documents:

• Personal contact details: phone numbers, email addresses and signatures
• Financial information: bank details and account numbers
• Sensitive identifiers: Medicare numbers and driver's licences
• Personal circumstances: medical information and financial difficulties

Manual redaction is slow and error-prone. A single missed phone number can lead to a privacy complaint.

AI redaction capabilities

• Pattern recognition: structured data such as phone numbers and emails
• Named entity recognition: names, addresses and organisations
• Contextual analysis: understanding when information should be protected
• Image processing: signatures and handwritten content

Structured data is the easiest to detect. Names, signatures and context-dependent information are harder, which is why confidence scoring matters. It lets the system flag uncertain detections for human review.

Best practices

• Tiered review: high-confidence redactions proceed automatically, and uncertain cases go to a person
• Audit trails: keep records of what was redacted and why
• Regular testing: check redaction performance periodically
• Staff training: make sure reviewers understand their role

Our detailed guide covers this topic: AI redaction for planning documents.

Compliance checking

AI can help with preliminary compliance assessment and surface issues early in the process.

Pre-lodgement screening

Before detailed assessment, AI can check:

• Completeness: are the required documents present?
• Consistency: do details match across documents?
• Zoning: does the proposed use align with the zoning?
• Development standards: are key measures within acceptable ranges?

This screening catches obvious issues that would otherwise delay assessment.

Assessment support

During assessment, AI can:

• Highlight relevant controls: surface the applicable development standards
• Flag potential conflicts: point out aspects that need closer examination
• Suggest conditions: recommend standard conditions for the application type
• Generate summaries: compile key information for the assessor

Limitations

• It cannot replace professional judgment: complex applications need human assessment
• Training data matters: performance depends on the quality of the examples
• Regulations change: systems must be updated when the rules change
• Decisions must be explainable: they have to be transparent and defensible

Communication and correspondence

AI can streamline communication with applicants and other stakeholders.

Automated status updates

• Generate notifications when milestones are reached
• Answer common questions about application progress
• Alert applicants to actions they need to take

Correspondence generation

• Draft request-for-information letters
• Generate acknowledgment responses
• Create determination letters from assessment outcomes
• Personalise templates with application details

Quality assurance

AI-generated communications should always be reviewed before they are sent. They should keep an appropriate tone for government correspondence, include clear contact details for human help and meet accessibility requirements.

Implementation roadmap

Councils considering AI should plan in stages.

Phase 1: Foundation

• Assess current processes and pain points
• Identify high-value uses for AI
• Evaluate vendor options
• Develop a governance framework

Phase 2: Pilot

• Introduce AI for one or two use cases
• Set measurement baselines
• Train staff on the new process
• Gather feedback and refine

Phase 3: Expansion

• Extend AI to more uses
• Integrate with core systems
• Optimise based on what you learned
• Document procedures and train more broadly

Phase 4: Optimisation

• Monitor performance measures
• Update training data and models
• Expand capability as the technology matures
• Share what you learn with the sector

Governance and risk management

Transparency

• Document how AI is used in each process
• Explain AI involvement to affected parties
• Keep human accountability for decisions
• Make sure decisions can be explained

Bias and fairness

• Test for unintended bias in AI outputs
• Monitor for unequal impacts
• Audit AI performance regularly
• Address issues promptly

Data security

• Protect training data appropriately
• Secure AI systems against manipulation
• Meet data sovereignty requirements
• Follow the relevant privacy frameworks

Change management

• Involve staff throughout the rollout
• Address concerns about AI and jobs
• Emphasise support for staff, not replacement
• Give staff clear escalation paths

The future of AI in planning

AI capability will keep evolving. Councils should watch developments in:

• Natural language processing: better understanding of text documents
• Computer vision: better analysis of plans and images
• Predictive analytics: forecasting application outcomes and timeframes
• Conversational AI: more capable interaction with applicants

The principle stays the same: AI should support human expertise, not replace professional judgment in planning decisions.

Getting started

1. Identify specific pain points that AI might address
2. Quantify current costs in time and resources
3. Research available solutions and vendor capability
4. Engage stakeholders, including staff, councillors and the community
5. Start small with a well-defined pilot
6. Measure outcomes and iterate

With careful implementation, NSW councils can use these capabilities while keeping the human judgment that good planning needs. To discuss a pilot, contact our team.