I am rebuilding my own firm around AI and writing down what happens, including the parts that did not work.

This is the talk version of that. It is specific, it has numbers in it, and it does not end with a prediction about the future of work.

The shape of it

Agents carry the admin. People keep the judgement. Every win and every failure gets written down where anyone can read it, which is a harder discipline than it sounds when the failure is yours.

What the talk covers

• What actually got automated, and the much longer list of things that did not survive contact with reality
• What it cost, in money and in the months where the rebuild competed with paid work
• The failure modes, including automation that quietly stopped running and nobody noticed for days
• Where the judgement stayed human, and how you decide that line rather than discovering it after an embarrassment
• What a small firm can do that a large one cannot, which is most of the point

Who it lands with

Owners and partners of professional firms, operations leaders, and anyone who has been asked to produce an AI strategy and suspects that the honest version is harder than the slide deck.

It is deliberately not a vendor talk. There is nothing to buy at the end of it.

Format

• 45 to 60 minutes
• Conferences, industry associations, partner retreats and founder communities
• Works well paired with a workshop the following day, while the room is still honest