How a pharmaceutical giant built a central nervous system for competitive intelligence

Intelligence scattered across a 22,000-person organisation

Servier is France's leading independent pharmaceutical laboratory: 22,000 people, €6 billion in revenue, present in 140 countries, with 20% of revenue reinvested in R&D and partnerships with institutions such as INSERM and Institut Curie. Anticipating scientific and competitive shifts is not useful there. It is the business.

The intelligence to do it already existed. It was simply not connected.

• Scattered by design, siloed by default. Departments ran separate systems: SharePoint repositories, email newsletters, individual research folders. Each team had its own approach, its own tools, its own audience.
• Insights that never crossed a boundary. A breakthrough in oncology research can change a partnership decision. Cross-functional sharing rarely happened, so geopolitical and market signals did not always reach the people making the call.
• Gathering crowded out analysing. Specialists spent their time collecting raw information rather than interpreting it. Teams re-entered the same research loops. The organisation accumulated data instead of producing insight.
• No consolidated view for leadership. Senior leadership saw fragments from individual departments, never a coherent picture that could orient corporate strategy.

Competitive intelligence is naturally dispersed across departments, and that's fine, because who better than the subject experts to set up precise, useful monitoring? What we wanted was to put them all in orbit around a central entity to encourage sharing.

One hub, not one more tool

Servier did not replace how teams research. It connected them. Elium became the point where human-curated intelligence converges, with each team keeping its own approach and contributing what it finds to a shared base.

• Aggregation, not centralised control. 25 active contributors tag, categorise and cross-reference content across roughly 15 specialised but shared spaces. Duplicate signals are welcome: two teams flagging the same thing is a measure of its importance, not waste.
• Distribution that builds a habit. Weekly digests keep 1,000 users engaged, and usage peaks every Monday as teams catch up on summaries matched to their interests. Thirteen tag categories keep material discoverable across therapeutic areas and functions.
• A direct line from the field to the executive committee. A monthly editorial committee turns platform content into executive briefings, so intelligence collected by specialists on the ground reaches quarterly strategy discussions instead of a departmental report nobody opens.

We went from intelligence scattered across departments to a unified hub that feeds our executive committee every quarter. It's like having a central nervous system for competitive intelligence.

The moment it proved its value

The test came at the J.P. Morgan Healthcare Conference, the most important annual event in the industry. Servier's team shared intelligence through Elium in real time during the event, putting competitive signals in front of senior management as they emerged. That settled the internal argument in the best possible way: leadership saw what the platform was worth during the most sensitive week of the year.

What began with 300 users focused on oncology now supports 1,000 across every therapeutic area in the group, more than a threefold expansion. Engagement rose as users recognised the quality of human curation: the platform grew because it was useful, not because it was imposed. The base holds more than 9,000 intelligence assets, a body of curated insight no single department could have built alone.

The next phase is applying AI to that corpus. The analyst's role does not disappear there, it grows: check that the information is true, despite persistent AI hallucinations, and add the strategic nuance that shapes a decision.

We have enough material to identify trends and generate insights we couldn't see before. The next phase we've started around AI is to involve reliable AI analysis of our knowledge corpus while maintaining the quality of human curation.

Intelligence is worth what it reaches, not what it collects. Servier left 25 contributors monitoring their own fields and built the one path that carries what they find to the executive committee.