Bio-pharmaceutical giant Sanofi is using its Snowflake data foundations to build a series of agentic AI services that help professionals around the company to work more effectively and save money that can directed to life-changing research and development.
Emmanuel Frenehard, Chief Digital Officer at Sanofi, says his company unified its data on Snowflake and then paired the data platform with the tech firm’s agentic AI services. The result is a new operating model that allows agents to run directly on Snowflake, meaning business professionals can eliminate manual processes and boost efficiency:
The data lake has now become far more transactional. The data lake used to be a place of reporting for us, but it’s becoming a place of intelligence and transactions.
Sanofi has been a Snowflake user for more than five years. The digital team looked at the organization’s data footprint and recognized they had a fragmented environment that relied on a complex myriad of tools. Their aim was to create a technology-enabled approach that would provide deep, cross-business information in a secure manner.
The first stage was the construction of a series of mini data lakes. Then Frenehard’s team began creating semantic layers across the data, so they could understand the value of its assets. They also implemented governance through a data ownership process, where professionals were given ground rules on how much data to keep and for how long.
One significant challenge during this transformation process was dealing with Sanofi’s huge variety of data types, including structured text-based documents and a plethora of un-structured information, such as PDFs, biopsy results, and blood samples. Frenehard says Sanofi worked with Snowflake to create data lakes of content.
While this approach helped Sanofi to see more of the value in its treasure trove of data, Frenehard says there were still limitations for professionals who wanted to produce deeper insights. People could only unlock significant value by running business intelligence dashboards on top of the data lakes.
However, Frenehard and his team has seen a step change by exploiting Snowflake’s AI capabilities. By working with Elementum, one of Snowflake’s technology partners, which specializes in building data-driven workflows that operate directly on top of the Snowflake AI Data Cloud, Sanofi professionals can create much richer insights:
Now the layer has become much more intelligent. Snowflake is no longer a compute and a storage capability; it’s also an AI capability. With Elementum, we built AI workflows straight onto the data lake, and there’s not many people doing that.
Today, Frenehard says Sanofi professionals can use AI-powered agents to query information directly in the firm’s data lake. In short, thousands of the firm’s employees can use Snowflake agents to receive insights and act.
While some organizations are exploring how agents might be inserted into existing business processes, Sanofi is going a step further and is using agentic AI in production every day to save the business money, which has life-changing implications because more cash is available for pioneering research. Frenehard says there’s more to come:
It’s an interesting approach because for the first time, we’re going to be able to create custom tailored business processes. We’ll be doing that without having to use our application layer, and by just using agents and data instead.
Sanofi is scaling AI across 80 countries, using Snowflake as the global data foundation for governance, collaboration, and production deployment. Most significantly of all, the organization has built an AI-powered Concierge service to help its staff exploit data assets and share knowledge.
This bespoke application for mobile and web is powered by Snowflake’s AI technology. The digital team’s stack draws on the personal agent CoWork, Elementum workflows, and a series of AI models, including the Claude family, such as Haiku, Sonnet and Opus.
Frenehard says Concierge and its models are fed knowledge about Sanofi. The service understands roles, geographies, cost centers, reporting lines, and the company’s broader policies and procedures. Applications include replacing contract management software with agents and using the technology for HR processes, IT support, and sales tasks.
As an example, all procurement orders, invoices, quotes, and RFPs are held in Snowflake. Professionals can run agents to analyze what the firm calls shadow spend – investment in the wrong areas with the wrong vendors. By using AI, traditional systems of record are bypassed, and agents act directly on information held in the Snowflake platform. Frenehard says the impact of agents on procurement and other areas is considerable:
We spend €18 billion a year buying stuff. Imagine just half a point of that spend optimized through AI. We also think that IT support can be managed through Concierge. We think that 80% of the problems that a user faces can be solved through AI, looking at history, and looking at knowledge articles.
Frenehard says about 65,000 of the firm’s 75,000 employees already use Concierge monthly. In the future, he expects the technology to be embedded across business operations:
As a member of staff, you will not be interacting with an enterprise application system; you will be interacting with an agent.
While the potential power of Sanofi’s data-enabled approach is considerable, Frenehard recognizes the journey from fragmentation to connected workflows has involved challenges. One of the key issues to overcome was that Sanofi was aiming to develop a new way of working, not just internally, but also in terms of interacting with Snowflake technology:
The DNA of Snowflake was not to write into it, it was a read environment. We’re writing back to it. When you say, ‘I want to place an order,’ your request doesn’t go into a system of record, it goes straight into Snowflake. So, we worked on writing data at scale inside Snowflake, which is a different way of operating.
Frenehard says working closely with Snowflake executives helped surface some of these issues and he advises other digital leaders to build similar bonds:
I built a rapport with the Snowflake CEO Sridhar Ramaswamy and some of his leaders, because we faced complexities. What we’re doing, I don’t think many other businesses have done that kind of work at scale.
Image credit – Pixabay
We’ll send you our top articles (and no marketing spam), no more than once a week.
Browse our Executive Intelligence library and hear from industry thought leaders on a wide range of agenda-setting executive topics.
Listen to more episodes at https://diginomica.podbean.com/
diginomica and the diginomica logo are trademarks of diginomica Limited.
© Diginomica Limited and its licensors 2013- 2026
Developed by BRAINSUM .
Leave a Reply