Skip to content
All work

PropertyPermit conditionsLand development

Planning permits turned into a condition tracker automatically

89%

full-condition accuracy

Status
In production
Built with
Document AIVision modelsStructured output

01The problem

Ten to forty planning permits a month, each ten to forty pages long, each carrying conditions somebody had to find, read and act on. A missed condition turns into rework, delay and a difficult conversation with a council.

02What we built

A document pipeline that reads the permit, separates the conditions from the surrounding text, and writes them into a structured tracker with the source page recorded against each one.

03The result

95% heading accuracy and 89% full-condition accuracy: roughly nine permits in ten needed no human correction. Planners use it to manage permit conditions and council correspondence.

Something like this in your business?

Tell us how the business runs and where the time goes. We’ll tell you honestly what’s worth building, and what isn’t.