An evening session we hosted · Sydney
How to build Agentic AI into your business
From personal use to organisational impact.
Most companies think they have adopted AI because their people use it every day. That saves one person twenty minutes. It does not change how the business runs. On a Tuesday evening in Sydney, Vivek Satasiya, our Chief Product Officer, walked a room through the gap between the two: a practical four-stage path from everyday AI use to measurable business results, with people kept in control.
About the session
UTS Startups is the startup hub of the University of Technology Sydney, and it gave us its event space on Broadway for a free two-hour evening session. Juliette O'Brien of OnlyFacts, who has worked with us for some time, helped bring it together.
Session page on Luma
Four stages, one employee
We told the whole path through one employee, Harry. First he drafts proposals faster with AI: useful, but invisible to the business. Then his team agrees one method, logs it and measures a real result. Then comes a first AI co-worker for one chosen task. Last comes governed operations: checked weekly, reviewed monthly, then moved forward, held or pulled back.
Pick a first task you can prove
A good first task for an AI co-worker can be checked, can be undone, and happens often enough to prove itself. The controls are written before anything is built: what it is allowed to do, when it hands over to a person, and how to roll it back.
Usage is not adoption
People using AI is usage. Adoption is when the process itself changes: it runs differently, its quality is controlled, and a business number moves. The method can be repeated, every decision and approval can be traced, and the impact is measured from start to finish.
The questions were the best part
The room mixed students, business owners and consultants who advise government on AI policy, and it did not let us stay theoretical. People brought the problems they face with AI in their own businesses, and many stayed after to keep talking. Vivek left them one question: where is your organisation stuck, at personal use, shared team standards, or your first AI co-worker?
Choose the first task for provability, not ambition.
From our session at UTS Startups, Sydney
