AI transformation is part of the leadership job.
At a glance
- Challenge
- A 90,000-person company had committed to becoming the first biopharma powered by data and AI at scale, and its leaders could not yet judge what that meant for their own businesses.
- Andrew’s role
- Co-led Drive Digital for Executives with Emmanuel Frenehard, Sanofi’s Chief Digital Officer, owning the people and capability side: I chose the partners, built it with my team and sat in it as a participant.
- Scale / context
- The top 150 first, then a cascade designed to reach the executive population of a global enterprise.
- What changed
- Twenty-four value theories from the first cohorts, ideas that reached the executive committee, and a cascade designed to reach more than 1,500 executives, which ran after I left.
Sanofi’s AI ambition required new leadership capability
In 2021 Sanofi’s leadership team asked itself an important question: did it have the skills and capabilities in data and AI to lead its own digital transformation? The ambition was already public, to become the first biopharma powered by data and AI at scale, and the honest answer was no.
I co-led Sanofi’s answer for its executive population, Drive Digital for Executives, with Emmanuel Frenehard, the Chief Digital Officer, and owned the people and capability side. I designed the approach, top to bottom: AI and digital capability for the whole organisation, starting with the top 150. I chose the partners, led the team that built and ran the programme, secured members of the executive committee as sponsors of the cohorts, and took a place in a cohort myself.
The programme was for the people who run the business, and it set out to give them what a transformation needs from them: a common language for data and AI, so that strategy and execution could be aligned; working relationships across functions; ownership of the transformation in the businesses rather than in a digital function; and the judgement to sponsor and fund the right work. Executives do not need to build models. They need to know which of the things in front of them is worth doing, and that judgement is built by deciding things. So the programme was built around decisions: each leader’s own idea, worked up in their own business, and carried to people who could fund it.
Executives learned by applying AI to their own businesses
I chose two partners for two different contributions. HEC Paris brought the teaching and the academic direction. OAO brought the applied side, coaching the teams on their own problems. The programme was for the top 150, in four cohorts of 30 to 40, small enough for everyone to contribute.
The design asked every executive to apply what they were learning to their own business. Six weeks of virtual work, during which each of them put forward two ideas for how data and AI could create value in their own part of it. Then three days together at HEC Paris, where the ideas were cut down to six a cohort and teams formed around them from across the cohort. Then four weeks working the surviving ideas up on the job, with coaching, alongside the day job rather than instead of it, into value theories: proposed applications of data and AI in the business, each with an explicit hypothesis about the value it would create. Then a pitch to a jury, and the winning idea from each cohort carried by its executive sponsor to a member of the executive committee to decide what happened next.
- Six weeks of theoryvirtual; two ideas from every executive
- Three days at HEC Parissix ideas a cohort; teams formed
- Four weeks on the jobwith coaching, alongside the day job
- The juryone idea a cohort
- The executive committeea decision, with resources or without
Ideas cut to six
Members of the executive committee sponsored the cohorts themselves. Their visible sponsorship told the executive population how much the programme mattered, and set the expectation for the ideas that would reach them.
I sat in it as a participant. Not observing from the side of the room: in a cohort, doing the work, pitching. It gave me first-hand experience of the demands, the learning and where the design could improve, and it answered the question a senior audience reasonably asks about a mandatory programme, which is whether the person who commissioned it thinks they need it too.
The strongest ideas went to the executive committee for a decision
What made this more than a course was where the output went. The value theories had a route: a team behind each one, a jury, and the winners in front of the executive committee for decisions on sponsorship, resources and next steps.
Twenty-four came out of the first cohorts, spread across the whole value chain, from research and early development through manufacturing and supply to the commercial and enabling functions. Blockbuster Booster came out of one of them: an R&D idea that uses data and AI to predict where Sanofi’s next blockbuster medicines will come from, and, once presented to the executive committee, it became part of Sanofi’s R&D strategic planning with dedicated resources. Work already under way gained from the programme too. The Turing project on the commercial side, which connects the right customers with the right physicians, received central visibility and a boost straight after the training.
A route to executive decisions gave the teams a practical reason to test their assumptions and strengthen the business case before presenting it.
Leaders had to judge the risks as well as the opportunities
A leader who can find an AI use case can also start one, and in a biopharma that is a question of data, risk, human accountability and controls as much as of value. The judgement the programme built has to weigh those together: the opportunity alongside the data it needs, the risk it carries, who is accountable for what it does, and the controls it runs under. Sanofi’s published Responsible AI Guiding Principles are the company’s framework for that, from fair and ethical through to accountable to outcomes, with a commitment to train every employee in responsible AI. They are the rules of engagement a capability programme of this kind needs beside it, so that leaders learn to see the opportunity and the terms of responsible use together.
We started with the top 150 and designed for the whole organisation
24
value theories from the first four cohorts, each worked up in the executive’s own business, pitched to a jury, the winners taken to the executive committee, as HEC Paris reports it
1,500+
executives the cascade was designed to reach across 2024 to 2026, in HEC Paris’s account of it
90+
net promoter score from the executives who went through the first four cohorts, in the programme’s own measurement
87%
of the top 150 who went through Drive Digital would recommend it to their peers, in the World Economic Forum’s 2024 write-up of Sanofi’s Lighthouse
93%
anticipated a sustainable impact on how they saw Sanofi’s data-driven transformation, in the same write-up
The top 150 were the starting point for the wider capability programme. The executive population came first because they sponsor everything beneath them, and the design was built to cascade from there.
I designed the cascade and put it to the executive committee in March 2024: a shorter programme, larger cohorts and regional events in place of the trip to Paris, keeping what had made the first programme work: each leader’s own application, work in peer teams and executive sponsorship. It ran after I left that September. HEC Paris’s account describes it as designed to reach more than 1,500 executives across 2024, 2025 and 2026, with six weeks of theory and three of applied work, an ideation workshop, a pitch to the executive committee and forty ideas a cohort. Those are HEC Paris’s figures for a programme Sanofi ran; the design was the one I put to the executive committee.
Two other parts of the design put AI into everyday work rather than into a programme. From January 2024, senior leaders were expected to include an AI goal in their objectives, so using it was part of the job rather than an interest. And plai, the company’s own decision-intelligence app, built with Aily Labs, put data and AI in front of thousands of decision makers in the flow of their decisions. Sanofi’s enterprise-wide digital upskilling, Democratizing Digital and Data, was recognised by the World Economic Forum as a Skills-first Lighthouse in 2024, and the Forum’s write-up records that 87 per cent of the top 150 who went through Drive Digital would recommend it to their peers and 93 per cent anticipated a sustainable impact on how they saw the company’s data-driven transformation.
Learning, sponsorship and accountability worked together
Guiding principles
You cannot delegate your understanding of AI. Sanofi’s executives did not need to build models. They needed to judge a proposal, sponsor the work and fund the right thing, and that comes from having made one themselves.
Put the learning in the work, and give the work somewhere to go. Every idea in Drive Digital was about the executive’s own business, and the best of them went to the executive committee. Connecting the learning to a real decision gave the work consequence, and gave the executives a reason to get it right.
Combine academic depth with practical application. HEC Paris taught the ideas. OAO coached the teams as they applied them to their own problems. Each did what it was best at, and the combination is what made the learning hold.
Build responsible use into the learning. The same leaders who can find an AI use case can start one, so responsible use has to be learned alongside the capability. In a biopharma the rules are not optional, and Sanofi has published its own.
The sponsor should take the programme. Members of the executive committee sponsored the cohorts, and I went through one as a participant, doing the work and pitching. It gave me first-hand experience of the demands, the learning and where the design could improve, and it showed the executive population how seriously the programme was meant.
Design the first cohort so it can be cascaded. The top 150 were the start of a design for the whole organisation. The 2024 cascade kept the shape, an idea of your own and a route to a decision, and changed the scale. Scaling needs a design that others can deliver consistently, rather than one that depends on the best facilitators being in the room.
Leadership capability must keep pace with AI
Since the first cohort went through in 2023, generative AI has put capable tools within reach of most executives, and what the leadership job asks has moved with it. The question Sanofi’s leadership asked in 2021 is one many boards are now asking, and the answer has the same shape it had then: leaders need to understand the tools, apply them in the business they run, sponsor the work and take responsibility for the results.
That is what the programme asked of Sanofi’s top 150. It is part of the leadership job now, and it can be learned the same way: inside the work, with the tools, with responsible use built in, and with a decision at the end.