key stats
145
questions answered across 20 conversations
10 consecutive minutes
of live conversation with zero technical failures
Bob Greenberg spent decades building R/GA into one of the most influential agencies of the internet age. But as the agency approached its 50th anniversary, R/GA faced questions that no org chart or manifesto could answer: how do you preserve the essence of a founder? How do you make institutional DNA conversational, searchable and alive?
In this discussion, R/GA Executive Creative Director Eli Mavros and Chief Technology Officer Nick Coronges discuss “Forever Founder,” a real-time AI avatar of Greenberg built on the Runway API with Characters. The avatar, which was built in partnership with Runway’s Forward Deployed Creatives, debuted at a recent party celebrating the launch of “R/GA by Design,” a new book chronicling the company’s history.
Tell us about R/GA and how generative AI fits into the studio's work today.
On the video side, we use generative AI daily across a multitude of clients. A lot of it is for pre-vis work or pitching ideas. It’s very much a tool we use to sell through ideas, get sign-off throughout the chain of command and bring concepts to life more robustly than we've been able to in the past. We're no longer making traditional ripomatic anthem films when we're pitching. We're now bringing TVC ideas to life using generative AI. And we're starting to use generative image creation and graphics across the whole agency, not just in the studio.
We also work directly with clients to integrate new technology. Our tech team is slightly different from a traditional engineering org. A lot of the talent are crossover, both creative and technologist. That's really where R/GA's sweet spot is: figuring out how to introduce these technologies in a way that helps clients with branding, communications and new services.
What was the genesis of Forever Founder?
We were originally going to do this experience two years ago. At the time, we'd done some prototypes for a town hall using another company's avatar tech, and the quality wasn't there. It wasn't very lifelike, just robotic. Now, R/GA is coming up on its 50th anniversary, and the book had been in the works for years, so we brought the idea back up.
Thanks to Runway, we were able to set up a prototype for Bob to see, and he was impressed with the fidelity. He said, “this feels right.” I think with the release of the book, he's thinking about his legacy. And the experience is so much richer than it would have been two years ago.
What made Runway the right technology for this?
There are two potential outcomes with a project like this. One is: “We think we can pull it off, but it looks like a lot of work.” The other is: “Wow, this works way better than we thought after the first few tries.” With Runway, we got 90% of the way there very, very quickly. Then it was incremental: tweaking the prompts, retraining the voice, a few passes on the knowledge base.
We collaborated directly with the Runway team pretty much every day, sharing what we were seeing and jumping on calls when things came up. Runway’s SDK made the API easy to use, but just being able to work with the team is what made it possible, because on our side, there was basically one developer working on this.
What happened at the actual launch event?
Everyone who interfaced with it was blown away. One moment in particular stood out: Bob was very close with his brother, Dick, who passed away in 2017. Dick’s son came up and spoke to the avatar, and he asked, “Who’s your favorite family member?”
The model responded immediately. It said: “Well, that's easy. Obviously it's my brother Dick.” And then it said this really beautiful thing about Dick being such a big part of the company and how close the two of them were. It was such a genuinely human moment, we all teared up. Because Bob talked about his brother in those terms throughout the book, the model understood that that would be his answer.
What does Forever Founder become from here?
The current demo is a prototype, but the long-term vision is for this to live online as something R/GAers can access from anywhere – across our global offices, with language versioning. Eventually, we want it trained on everything R/GA: all of the company history and data, not just the book. Imagine asking it about a Nike case study and having it pull up and play that case study for you. It could be a second-screen experience or an onboarding tool for new hires.
What makes this particular data set so powerful is that Bob had a really strong hand in crafting the book. He focused his energy on it after stepping down as CEO. So the model isn't trained on a historian's account of R/GA. It's trained on the story through Bob's lens, in his voice. That changes what it is fundamentally. And because of generative AI, those principles and that essence can be translated into future situations that we can't even predict right now. The book doesn't have to die the moment it gets printed.
