WorkAI unit — WPP Production France
AI unit — WPP Production France
Moving generative AI from experimentation to an offer an agency knows how to sell, price and deliver — within the constraints of a large group.
The context
My current mission at WPP Production France: setting up the agency’s AI unit. Brand requests come in fast — videos, content, variations — with short deadlines and often vague expectations about what AI can really do. The agency knew how to produce. It had to learn how to sell, price and deliver projects whose production methods are still new, within a group that has its own tools and its own rules.
My role
I am structuring the unit end to end: pre-sales and the way AI projects are presented to brands, pricing models, how production is organised between in-house teams and freelancers, and delivery constraints. I work on projects for brands such as Schneider, Johnson & Johnson, Sanofi, Accor and Carrier.
What it changed
The biggest change is in pricing. An AI project is not sold like a traditional production: you do not bill shooting days, you bill iterations. I built a model that reads clearly for the client — clear units of work, a storyboard frame, a scene, a set number of revision rounds per deliverable — and folded AI-specific costs, licences and usage, into technical fees. Negotiation now focuses on specific points rather than a lump sum, and the agency can answer a request received on a Friday evening with a proposal on Monday. In pre-sales, interactive moodboards attached to the proposal let clients picture the result before committing.
What was delivered
— A pre-sales pitch and format for AI projects, with interactive moodboards attached to proposals — Pricing models by profile and by unit of production, AI costs included — A production organisation combining in-house teams and freelancers, with visibility on everyone’s workload — A delivery framework: revision rounds, approvals, managing expectations
What remains
AI does not only change how things are produced, it changes how they are sold. As long as an agency prices an AI project like a shoot, it loses either its margin or its client. It is also the project that ties my whole path together: as with virtual reality ten years ago, the technology arrives before the uses, and the job is to make it reliable, sellable and deliverable.