Sanofi AI Strategy: In-House AI at Scale, Under a New CEO

Sanofi set out to become “the first biopharma powered by AI at scale” and built much of its AI in-house. About 80% of employees use its own AI assistant, and agents run on its own data across the business. In February 2026, its board replaced CEO Paul Hudson, the executive most associated with that ambition. AI did not feature in new CEO Belén Garijo’s first quarterly results statement in October 2026.

The ambition is AI at scale. Chief Digital Officer Emmanuel Frenehard said in June 2026: “This is how Sanofi becomes the first biopharma powered by AI at scale” (Snowflake). It is one of the clearest bids in pharma to become one of the AI native organizations.

AI is framed as infrastructure. Hudson wrote in February 2026 that AI “has shifted from experimentation to becoming a vital part of our infrastructure” (Fortune). Frenehard put it this way at VivaTech in June 2026: “AI means reinvention, not automation. When it amplifies our people, we can move faster” (translated from French, Sanofi).

The focus spans the business. Sanofi uses AI across:

• R&D: drug development committee meetings begin with an AI agent’s assessment of whether a drug should advance, plus AI literature review and in silico trial modeling
• manufacturing and supply: AI that predicts low-inventory risks, speeds the transfer of new medicines into production and improves equipment reliability
• commercial: Concierge for Field, an AI agent that prepares sales reps for doctor meetings
• enterprise functions: agents in IT, HR and procurement

Sanofi has not said what it will not prioritize.

Sanofi built instead of buying. Frenehard initially turned down an enterprise ChatGPT license and did not widely roll out Microsoft Copilot after a small pilot: “We looked at it, and said, ‘This is lame’” (Fortune, May 2026). Instead, Sanofi built Concierge, an internally hosted generative AI companion launched in October 2024:

• It is used by 60,000 employees, about 80% of the workforce. Most non-users work in manufacturing.
• It draws on HR, IT, policy and organizational data, and connects to agentic systems.
• Sanofi says more than 70,000 employees are now supported by AI in their work (June 2026).

Its own data is the foundation. Agents run on a central data lake built with Snowflake, and in June 2026 Sanofi said it is deploying agents across R&D, procurement, IT, HR and field sales (Snowflake). It also has an internal development platform, AI Foundry.

Partners serve specific jobs:

• OpenAI and Formation Bio: custom AI software for drug development (May 2024)
• Anthropic: Claude for Healthcare (May 2026)
• Aily Labs: plai, the decision-support app rolled out at scale in 2023

Adoption started small and spread. In 2024, Hudson took about 15 employees with no AI expertise into a hotel nightclub to experiment with plai, using a Fight Club-style rule to build curiosity. plai users rose from 15,000 to 18,000, and Sanofi ran its first company-wide hackathon (Fortune, November 2024). It was an early, unusual form of AI champion programs. In April 2026, a company-wide “digital month” included Concierge training.

Leadership changed in 2026:

• The board decided not to renew Hudson’s mandate. He left on 17 February 2026 (Sanofi).
• Belén Garijo, former CEO of Merck KGaA, became CEO on 29 April 2026. Sanofi said she would bring “increased rigour” to executing its strategy, with priorities in productivity, governance and R&D innovation (pharmaphorum).
• Emmanuel Frenehard, Chief Digital Officer since 2023, is not on the new Executive Committee from September 2026, but still reports directly to Garijo (Sanofi).

The workforce message is vendors, not jobs. Frenehard warned that announcing AI-driven job cuts would “create more friction,” and said IT and procurement savings will come from external vendors, not employees (May 2026). Separately, Sanofi booked €308 million for redundancy plans in the first nine months of 2026, without linking them to AI (Q3 2026 results).

Governance has a name. RAISE (Responsible AI at Sanofi for Everyone) rests on five pillars: accountability, safety, fairness, transparency and eco-responsibility (June 2026). The sources reviewed describe the pillars but not how reviews work.

AI fluency, coded against the Lead with AI Fluency Matrix (September 2026 edition):

• Fluency floor: Level 1 (Chatting), evidenced by use. About 80% of employees use Concierge, the strongest usage evidence in the index. Use shows a habit, not a skill standard, and Sanofi has not published one.
• Highest level evidenced: Level 3 (Integrating), company-claimed. AI predicts supply risks, reviews literature and informs development decisions. Agents in IT, procurement and sales come with savings estimates, not results, so Level 4 is not evidenced.
• Portfolio design: In-house and broad. Most office staff use one internal assistant, and a central digital team builds agents on shared data. Manufacturing staff are largely outside it so far.

Sanofi reports results by use case. All figures below are company-claimed:

• Supply chain: AI helped avoid $300 million in revenue risk and predicts 80% of low-inventory risks before they happen (February 2026).
• R&D: AI helped find 10 new drug targets in one year (February 2026). Its AI literature review reaches more than 95% accuracy (June 2026).
• Portfolio decisions: Sanofi reallocated $800 million based on advice from its AI tool in 2023 (reported November 2024).

Some figures are estimates, not results. Frenehard estimates that agents resolving up to 80% of IT requests could save €10 million a year, and that Concierge for Field saves each sales rep a day a week (May 2026).

Two things are missing: a group AI value figure and AI targets. Most published results came from the former CEO, and Sanofi’s third-quarter 2026 results did not mention AI.

Change log

• June 2023: plai, built with Aily Labs, rolls out at scale, and Hudson sets the goal of becoming a pharma company powered by AI at scale.
• May 2024: Sanofi partners with OpenAI and Formation Bio on drug development software.
• October 2024: Concierge launches.
• November 2024: plai reaches 18,000 users, and Sanofi reports reallocating $800 million on AI advice in 2023.
• February 2026: Hudson calls AI “a vital part of our infrastructure”; two days later, Sanofi announces that the board will not renew his mandate.
• April 2026: Belén Garijo becomes CEO, and a “digital month” includes AI training.
• May 2026: Concierge reaches 60,000 users, about 80% of employees.
• June 2026: Sanofi announces agents across functions with Snowflake and presents RAISE and AI Foundry at VivaTech.
• September 2026: The new Executive Committee takes effect, without the Chief Digital Officer, who still reports to the CEO.
• October 2026: Third-quarter results do not mention AI and show €308 million of redundancy plans for the year to date.

• Disclosure: 4 of 4. Sanofi answers all four questions, though its measurement is by use case, with no group value figure.
• It built instead of buying. An in-house assistant used by about 80% of employees is the strongest usage evidence in the index.
• Its agents run on its own data, across R&D, supply, sales and enterprise functions.
• The numbers need care. Net sales rose 16% from 2022 to 2025. Most of the 16% fall in reported headcount is the Opella separation; excluding it, the fall is roughly 5%.
• The say-do gap is continuity. The CEO who made AI part of Sanofi’s identity is gone, the digital chief is off the Executive Committee, and AI was absent from the latest results.
• Watch Garijo’s first strategy update. The test is whether “AI at scale” survives as a stated priority, with results attached.

What the numbers show: Net sales grew 16% from 2022 to 2025. Reported headcount fell 16%, but most of that is the 2025 separation of Opella, the consumer health business with about 11,000 people; excluding it, the fall is roughly 5% (an estimate). Revenue per employee was about €570k in 2025.

Revenue, headcount and revenue per employee, FY2022 to FY2025

• FY2022: €37.65bn net sales, 91,573 employees including Opella
• FY2023: €37.82bn net sales, 87,994 employees including Opella
• FY2024: €41.08bn net sales, 84,587 employees including Opella
• FY2025: €43.63bn net sales, 76,493 employees, €570k per employee

Change, FY2022 to FY2025: net sales up 16% on a like-for-like basis. Headcount and revenue per employee are not comparable across 2025 because of the Opella separation.

Notes: Revenue is net sales, as reported. Headcount is at year-end. Net sales exclude Opella, the consumer health business, for all four years. Headcount includes Opella’s roughly 11,000 people until its separation on 30 April 2025, so revenue per employee is only shown for 2025, and the 2025 headcount drop mostly reflects the separation. Trend figures stay in € so exchange-rate moves do not distort them. At the 2025 average exchange rate, latest revenue per employee is about $644k.

These figures give context on scale and productivity. They do not show that AI caused any change. Sources: Sanofi 20-F filings.

What is Sanofi’s AI strategy?

Sanofi aims to be the first biopharma company powered by AI at scale. It builds its own AI tools on its own data, including Concierge, an AI assistant used by about 80% of employees, and agents across R&D, supply, sales and enterprise functions.

‍

Who leads AI at Sanofi?

Emmanuel Frenehard, Chief Digital Officer, leads digital and AI and reports directly to CEO Belén Garijo, who took over in April 2026. Former CEO Paul Hudson, who championed the AI strategy, left in February 2026.

‍

What is Sanofi’s Concierge?

Concierge is Sanofi’s internally built generative AI assistant, launched in October 2024. It draws on company data and systems and was used by about 60,000 employees by May 2026. A version for sales reps prepares them for meetings with doctors.

‍

Is Sanofi cutting jobs because of AI?

Sanofi has not linked job cuts to AI. Its digital chief says AI savings in IT and procurement will come from external vendors. Sanofi booked €308 million for redundancy plans in the first nine months of 2026, without tying them to AI.

‍

How is this profile put together?

Lead with AI profiles use public sources only, such as annual reports, earnings materials, filings and on-record interviews, dated at the time of writing. AI fluency is coded against the Lead with AI Fluency Matrix. Company figures are marked as company-claimed unless they have been independently verified.