Microsoft: Using AI Agents to Handle Clinical Data Surge
Medical communications agencies face mounting pressure from the volume of clinical data they must process. Cactus Life Sciences has responded by embedding Microsoft 365 and the Copilot AI platform into its operations.
The global agency employs more than 350 professionals with science degrees. They convert clinical and scientific research into insights for healthcare professionals, payers and patients.
Automation replaces manual processes
The company previously relied on manual document reviews and structured data extraction. These tasks were managed effectively but could not scale efficiently.
Microsoft deployed its 365 Copilot in phases to address operational constraints while protecting data integrity. Security was the priority because the agency handles proprietary pharmaceutical information.
The deployment required trusted authentication and dedicated project environments to protect client data. Microsoft applications allowed adoption within Cactus Life Sciences teams.
"We didn't just want to automate tasks, we wanted to reimagine how work gets done," says Odity Mukherjee, Lead AI Transformation at Cactus Life Sciences.
"With tools like Copilot, we've been able to rethink our workflows from the ground up, creating new efficiencies that free our teams to focus on what truly matters: delivering exceptional science to our clients."
Custom agents target specific functions
The company deployed more than 30 custom automation agents. These agents automate components of larger functions rather than replacing entire roles.
Writers can focus on analysis quality while automation manages repetitive data processing. The custom agents retrieve and structure information from scientific literature.
Additional automation tools support quality control tasks. These include checking abbreviations, ensuring uniform formatting and verifying compliance with regulatory and publishing standards.
The agents can extract and compare insights across multiple documents. According to Cactus Life Sciences, structured data extraction is now 35% to 50% faster than with previous workflows.
Training programmes support adoption
Project managers use the agents to summarise email threads and generate task lists. Scientific writers process large volumes of articles more efficiently.
The company launched a Copilot Champions programme to identify internal advocates who would encourage experimentation. A dedicated training team implemented modular education programmes to teach employees the capabilities and limitations of AI tools.
A centralised repository was created to share effective prompts, agent ideas and best practices: "We built our Copilot Champions programme about six to eight months ago to create a vibrant community," adds Odity.
"This helps drive experimentation and allows new ideas to emerge, whether it's developing agents or finding new ways AI and automation can augment our work." The agency maintains a human-centric approach to technology.
Processes remain unchanged to address concerns about AI use. Technology supports human professionals and all automated outputs are reviewed.
