François-Xavier Pierrel, group chief data and adtech officer at French TV network TF1, is an experienced executive with a background in tech-led innovation. At TF1, he’s helping his organisation embrace business transformation. Pierrel suggests the scale of change means his role covers a broad spectrum of activities from data to advertising technology.
“Most of my job is everything in between, especially in a company where transformation is at the heart of what we do,” he says, referring to the rise of online streaming and continued provision of linear TV scheduling. Maintaining both routes to the viewer is a lot of effort.
“We are developing everything through digital, but we need to protect everything in linear, so there is a kind of paradox in the approach,” he says.
“We need to protect a world that is absolutely not digitalised. And we need to accelerate in a world that is changing every day.”
Pierrel joined TF1 in October 2023. His appointment followed Rodolphe Belmer’s assumption of the CEO role in 2022. Belmer aims to transform the broadcaster into a digital streaming leader through ad-supported platforms, such as TF1+, and high-reach partnerships, including one with tech giant Netflix.
“What we do is make sure that we can technically, from a data standpoint, execute the vision of the company,” says Pierrel, emphasising his team’s focus. He says the data and adtech department was created to help the organisation exploit its information by building interoperability and developing an ecosystem that can help the business monetise its assets.
“We need to become a destination platform,” he says. “We need to create this habit among viewers, and this habit comes from better reach, better content and better user experiences, and to manage all this, data becomes a key element of the proposition.”
Shifting direction
Pierrel’s broad CV includes stints at US technology giants, including Microsoft and Meta, and French blue-chip enterprises, such as automotive firm Renault and advertising specialist JCDecaux. While he enjoyed his time in Big Tech between 2011 and 2017, Pierrel was eager to try something different.
“When you’re working for companies like Microsoft and Meta, you are on the locomotive; everybody wants to work with you,” he says.
“People often swallow whatever you say. But I was willing to do something else. I’d been working for Chinese, Taiwanese and American firms for 20 years. I said, ‘Okay, I’m going to take everything I learned to French companies that need to be transformed to evolve their business models.’”
Pierrel says shifting from a fast-moving, tech-first firm like Meta to more traditional enterprises wasn’t straightforward. Large-scale, data-powered business transformation requires high-level sponsorship from the executive committee. After a stint with Renault, he introduced data-led change at JCDecaux as group chief data officer.
“We had to build up everything,” he says. “I built the team and capacity. Then the results from this transformation came, and I got a taste for change. Now, I’m trying to do the same at TF1. It’s about how you leverage tech and people to transform companies. What’s interesting is how you leverage all these elements together to take the company on another path and how you help elevate the change.”
At TF1, Pierrel reports to deputy general manager and executive committee member François Pellissier. While some data chiefs have a tight reporting line to the chief technology officer, Pierrel believes it’s important to connect data to the lines of business, not just the IT organisation, especially in the artificial intelligence (AI) age, where the monetisation of information assets has never been so important.
“Doing IT is one job – doing data is another,” he says. “When you connect data in a tech ecosystem to the business, you create a more concrete approach. Rather than focusing on pipes and plumbing, we impact the company. We incentivise intellectually the work that we do and make it really concrete by impacting the numbers across the business.”
Extending capabilities
Looking back on his three years with TF1, Pierrel says one of the first things he recognised was the importance of strong underlying foundations for change.
“I felt the shift to digital at scale was not at the right level for the project being set to bring impact,” he says.
“So, the first thing was saying, ‘Guys, we need to clean the house’. We got rid of all the secondary technologies and the mix of bits and bytes across the organisation. We made some concrete choices.”
Rather than using multiple datasets and platforms, TF1 decided to implement the Snowflake AI Data Cloud across the organisation. The company already used the technology, but the pre-existing Snowflake platform was more like a data warehouse than a foundation for change. Pierrel says he wanted to do more with the technology’s capabilities.
“We wanted to create an operating system for data, something that is essential, and Snowflake was the heart of our ecosystem,” he says, adding that the first product the team built was its audience graph, the initial version of which was delivered in August 2024.
Known as Graph ID, the technology provides a structure for all TF1’s logged-in accounts. The broadcast giant has about 25 million profiles with around 100 criteria. Each profile includes key data points, such as name, email, device, connections and tastes.
“We said we are going to build up this audience graph that will be the same for everybody,” says Pierrel. “It will be a gold model for the company’s data. Whether you want to target ads or marketing automation to launch your new programme, it will be the same information, because in the end, the same streamer is exposed to ads and to programmes.”
Embracing agents
Pierrel says the Snowflake platform and Graph ID allowed his team to refine its data transformation efforts. Having built these underlying foundations through 2024, he then turned to internal culture.
“If you want to do agentic AI, data needs to be extremely clean, extremely organised, well governed, with strong lineage, and all the basics of data, which is what we did in 2024,” he says.
“In 2025, we reorganised the team to have product managers. So, we also changed the way we operate on a human scale. Now that we have these two approaches to data and people, we are in a position to ingest new technologies more easily.”
Pierrel says TF1 used Snowflake’s CoWork agentic service to develop its first AI agent. As part of this process, the company ran a hackathon. The nine-day event, which was run alongside Snowflake’s consultants, led to the development of the broadcaster’s agent for its sales team, the first version of which was released in late August 2025.
“The sales guys said, ‘How am I going to use it?’ he says, referring to the implementation of the agent. “We said to them, ‘You’ll prepare your meeting, and you’ll report. This agent is something that you didn’t have before. You will have more weapons. You will move from maybe a pure sales relationship to more consultative selling, and we’ll help you take the next step.’ So, that was the first version, and we kept improving.”
For example, Pierrel says a Teams integration means sales staff will be able to interact with agents and discuss sales data and plans.
“When you do ads, you need measurement, proof of performance, auditability and transparency,” he says. “Everything we do is going in that direction. We are supporting our team to have easy tools whenever they need them, with the highest grade of quality in the results.”
The team is also exploring how agents can help data analysts save time. Pierrel says the agentic technology will help to automate reports, as many as 80% of which involve pulling together similar information every morning.
“We want them to focus on tasks that really create value for the business,” he says.
Sharing lessons
Crucially, TF1 is actually exploiting agentic AI. While other data chiefs and CIOs might struggle to turn AI explorations into production services, Pierrel and his team are creating value through agentic AI. So, what are his lessons for other digital leaders considering agents? The key to success is getting as many people involved as possible.
“AI is buzzing everywhere,” he says. “If you have new ideas and use cases, come around the table and help us as data specialists to think about the next projects and products. Usually, technology ideas come from technicians and go to sales and marketing or business units. Here, agentic AI is an opening where we can help people work differently. Our first agents were a way for us to gain our medals.”
Pierrel says these successes have helped demonstrate the benefits of agentic AI. He expects more developments in this area soon. Today, he looks back on the changes that have already taken place and reflects on how the pace of change continues to quicken, both in terms of innovation and across internal roles and responsibilities.
“We started our transformation in early 2025, but it was on a pure data standpoint where we were operating with data engineers,” he says.
“Now, data engineers are changing, and they are becoming AI engineers – ops becomes AI ops. So, everything is moving. And I think what we are going to do in terms of the organisation is to keep adjusting all the time.”
In addition to its Snowflake agentic technology, Pierrel says TF1 uses Claude Code, which helps developers reduce the time to production by as much as 50%. He expects the roles of developers, engineers and data specialists to continue changing in response to innovations in emerging technology. The message for all digital leaders is to be prepared.
“Nothing will be monolithic anymore,” he says. “You will need to adjust all the time. Look at the rhythm and pace of innovation. As chief data officers, we need to ingest these transformations.”
