Nu Davidson said at the start of the session: “I don’t actually even want to start the journey because I’m just like, I actually can’t understand any of it.” She’s not an outlier. That sentence describes a significant portion of the people who would benefit most from this methodology.
She left 72 minutes later with a working agent on her machine, two AI-generated ad images approved in the session, a video ad, a complete Ads Manager build guide, and a clear picture of what to do before she clicks publish. The agent didn’t remove her judgment — it removed the parts that were blocking her from applying it.
Mack Brands taught us that the workshop surfaces the leverage. LiquidGold teaches us that with a tight brief and a pre-loaded agent, 72 minutes is enough. The format adapts. The outcome holds. One operator. One agent. Operational.
“So many of these companies are installing all of these AI tools, but they’re not installing it into the people.”
The agent sat in Nu’s folder. Not on a company server. Not in a shared IT system. On her machine, in her workspace, logged into her account. That’s what makes it stick. The install is into the person — her workflow, her account, her campaign. It doesn’t exist without her. That’s the whole idea.
Paul Domanski · In-Seat AI canonical statement · 19 May 2026