How I built an AI planning tribunal in a weekend by orchestrating AI agents

Description and use

Setback automates the filtering of resident objections to development applications, ensuring only legally relevant grounds are submitted while transparently explaining rejected points.

Problem

Residents submit objections to nearby development applications that include concerns not covered by NSW planning law, leading to ignored submissions and frustration.

Outcome

Setback automates the filtering of resident objections, ensuring only legally relevant grounds are submitted while transparently explaining rejected points, improving resident understanding and council efficiency.

Workflow

• Collect resident interview transcript and uploaded photos/plans
• Run AI reviewers to split objection into individual grounds
• Cross‑check each ground against NSW statutory list
• Generate plain‑English explanation for rejected grounds
• Submit valid grounds to council
• Return explanation to resident

Evidence

Setback runs a small panel of AI reviewers over a resident's interview transcript plus whatever photos or plans they upload, splits the objection into individual grounds, and checks each one against the actual statutory list before anything gets submitted anywhere.

Source

The two nastiest bugs of the build were both invisible to a fully green, several-hundred-test offline suite and only surfaced once I forced a genuinely deployed, end-to-end run.

Source