Why myLot’s answers are right, and how we know

Start with what the model is not asked to do

myLot’s AI never draws on a language model’s general knowledge of planning. Which controls apply to a site is resolved by code from your digitised planning scheme and authoritative mapping, not decided by the AI. Where a control is objective, a numeric standard or a spatial test, it is handled deterministically.

The AI’s job is to interpret the application against a fixed framework, and every finding carries the clause and the page it came from.

Grounded in the scheme, kept current

We digitise your council and state planning instruments, from whatever format they are in today, through an ingestion pipeline that turns PDF and Word into structured data with every clause traceable to its source. The result is held in a versioned library, and the same pipeline keeps it current: 95% of scheme and GIS updates go live within 48 hours.

Each control carries assessment notes that record how it is to be interpreted where the wording allows more than one reading, and your planning team can maintain those notes, along with application requirements, referral triggers and your conditions library.

Tested against your decisions

At onboarding we build your council a test suite of its own: scenarios across your development types, zones and overlays, with expected outputs stipulated by your planners and approved by your council. Those scenarios become regression tests, re-run after every model change, prompt change or scheme amendment.

Before any product goes live, it runs read-only against applications your council has already decided, and its output is compared with the RFIs and reports you actually issued.

Measured continuously

An automated evaluation layer compares what each AI process returns against what it should return given its inputs. When outputs drift, the process is isolated and re-run at volume to establish whether the deviation is systematic, and can be run across alternative models to compare accuracy before any change is made.

Metrics are reviewed weekly and prompt performance at least quarterly. In production, every officer accept or dismiss is logged, which gives a running measure of accuracy in your council.

Reviewed by planners

Our in-house planners write the instructions the AI works to alongside our engineers, review production outputs continuously and audit all council feedback. Every flagged output becomes a logged issue that a planner and an engineer work through, and the fix is traced back to the person who raised it.

What the numbers say

In testing on live applications at a large Queensland council, Check identified 100% of the issues the planning team raised in action notices and the majority of RFI issues raised by planners and specialists.

Where specialist teams had written specifications, Check identified every deviation from them. Across all councils, Enquire has answered over 16,000 enquiries with no complaints from applicants.