Overview of cases of use of generative artificial intelligence deployed in France by the World Group of Creative Industry.
International giant of creative industry, WPP seized generative artificial intelligence very early on. The company uses it throughout the life cycle of the creative process, from market analysis to the realization of finished products through the ideation and creation of supports to sell concepts to customers. User cases that may vary depending on the subsidiaries.
On the VML side, one of the main creative companies of the WPP group which has 600 employees in France and 30,000 worldwide, AI agents have been created for each customer in order to scan their brand, their audiences, their market. “The generative AI makes it possible to put the finger on important questions which are not necessarily formulated in the brief. Then, it allows us to extract from the web the feeling of Internet users concerning a brand. In terms of ideas, it is also a good way to identify those which have already been used and thus to orient our reflection on truly original and innovative axes”, describes Martin Iselt, Executive Creative Directive Paris, another creative agency of the WPP group, focused on brand identity. “The prompts also allow us to generate brainstormings that we then refine to find original ideas. It is an incredible time saving. The ideas that humans identify in two days, the machine takes a few seconds to find them.”
At Ogilvy, another creative agency of the WPP group, the ideation remains the pre-carré of the human. “We are convinced that the Gen AI does not replace human experience and intuition,” argues David Raichman, Executive Creative Director of Ogilvy Paris. “On the other hand, when ideas have emerged, AI will help us project them into graphic, videos, sound content in order to defend them in the face of our customers.” The creative teams thus avoid the long and tedious work of creating models via Photoshop while benefiting from a support to design realistic content. Same approach within Landor Paris. “At this level, there are many use cases: the generative AI will be able to complete an image, modify the background. It can also delete an object in a visual or generate a Story Board starting from a first sketch,” adds Martin Iselt. “It can also create visuals that are not present on image banks, for example that of a hand holding a hamburger. This saves precious time in retouching and graphics work, especially in our example on the rendering of the hamburger.”
A conceptualization of endless ideas
And Martin ISELT to add: “With AI, we no longer have a barrier in terms of conceptualization of ideas. Everything is now imaginable. Designers become creators of creation. It is up to them, in the end, to define among the images generated those which stick to the brief. Knowing that the basic mechanism remains the same. It is a question of identifying the right idea, then of generating one or more images to sell the project, and then defend the concept, customer.”
At Landor, we use GEN AI as a tool for titles for advertising. On the side of Ogilvy, it also serves to work on styles, in clothing for example. This allows customers to project themselves on a most concrete basis.
VML also uses virtual group focus starting from the audiences of its customers. “Instead of testing concepts of content, text, text, audio or video with real focus groups, we created thanks to the geni of synthetic group focus. This does not allow you to replace real focus group, but thanks to this technique, we limit the errors by identifying for example the words or the phrase turns that do not attract attention”, explains Vincent Druguet In France. “To carry out this task, we have developed our own technology based on the LLM on the market.”
“In the end, there is not a campaign that comes out of our home without a part created thanks to Gen Ai”
For the text, WPP uses widely in generative AI to design its textual content. On the side of VML in particular, the editors use in this perspective to the LLM of Openai, Gemini and Claude Fine trusted with agents. Objective: optimize the writing of posts published on social networks by providing AI from indications on the brand’s brand identity in terms of Wording (style, tone, hashtag …).
At Landor, we refrain on the other hand to go through the Gen AI to make the finished product, whether for the photo or the video. “Beyond the problem of image law, generative AI is not yet at the level in terms of resolution. It is also not capable of making certain details as is the case in food products or cosmetics. Ditto when we must integrate a human subject,” specifies Martin Iselt. “For us, AI therefore remains a source of inspiration and creation of insights which are then validated by the customer before the final shoot and the production which, themselves, remain currently carried out in real situations.”
On the side of VML, on the other hand, we do not hesitate, conversely, to resort to generative AI in the design of finished products. The agency uses Stable Diffusion, Adobe Firefly, as well as BRIA which is a platform trained on validated content in terms of copyright. The agency also works more and more with the tools of Gen Ai from the services of Getty Image and Shutterstock. “Each of these solutions with its own advantages, we are led to mix them. For example, we create a character in Midjourney, but we generate the face with Gen Ai from Getty Image insofar as it will be free but also more realistic,” adds Vincent Druguet at VML. “In the end, there is not a campaign coming out of our home without a part created thanks to the Gen Ai.”
The question of copyright
Product design, shooting, landscape generation … VML uses generative AI in many areas. “As soon as there are characters, however, ask the question of copyright, but also of the quality of the rendering,” recognizes Vincent Druguet. And David Raichman at Ogilvy added: “Gen Ai will find it difficult to capture the emotions of the actors and grasp a game of actors.” On this point, Vincent Druguet relativizes: “A solution like Pimento allows from a shoot of photos taken with a model then dismiss his image to generate new scenes which really give the impression of having been produced with the same person.” In terms of video, VML is just starting to test solutions like Runway. This is also the case of Heygen’s offer which generates video clones speaking for you, including in several languages. A lever that makes it possible to drastically lower costs by going beyond the traditional realization phase.
On the side of Ogilvy, we also do not refrain from making finished products by leaning against the generative AI. “Our artistic directors become practically directors by leaving an AI specialized team, baptized with us ia lab,” said David Raichman. “Starting from a Story Board, we can for example follow a generative process of image-to-video to reach the final film sought.” We will be over a chain of trades centered on realization, comedy, photography. A concession that all customers of these agencies are not yet ready to do.
