Why I started Operantra

Most small and mid-sized firms I meet in construction, property and professional services still run on drawings marked by hand, Excel sheets, paper registers and WhatsApp threads. Everyone tells them AI is the answer, and then hands them a chatbot on the side.

I think AI should be an operator inside the work itself. So I do one thing, the same way each time: I enter one workflow, learn exactly how the owner makes decisions, and rebuild it as a system they run every day. The AI does the reading, measuring and sorting. People keep the decisions that carry liability.

Measure reads an architect's drawing set sheet by sheet and builds the Bill of Quantities. It takes scope from the finish hatches rather than room labels, links every quantity to its drawing, and marks each one measured, counted or provisional. When an architect's comments come back, the revised bill is rebuilt the same day, and counting from the drawings has caught items that hand-made bills had missed.

Property OS took housing societies off the paper register. Amounts set by the committee are entered, never invented; every bill is numbered without gaps; mistakes are corrected by credit note; and residents get their bills on WhatsApp, where they already are.

Firm OS runs the daily work of a chartered accountancy firm. Documents that arrive on WhatsApp are read against a fixed schema and move through a pipeline per service, so there is one queue, and a clear owner for every next step.

I designed, built and deployed all three. The part I'm proudest of is that the owners now make their own changes to the systems they run. That's the point: AI built around how small firms actually work is practical today, not a large-enterprise luxury.