How Cactus Champions Microsoft’s AI Agents for Healthcare

The increasing pressure on modern healthcare has an AI cure with Microsoft tools. 

Cactus Life Sciences – global life sciences agency – has deployed Microsoft 365 Copilot and more than 30 custom automation agents to handle complex clinical data processing tasks.

The implementation addresses operational bottlenecks while meeting strict security requirements for pharmaceutical data.

Cactus Life Sciences processes clinical and scientific research for healthcare clients using a team of more than 350 professionals with advanced science degrees.

According to the agency, manual document reviews and structured data extraction were previously managed effectively but lacked scalability.

Microsoft deployed its 365 Copilot platform in phases to automate specific workflow components while preserving data integrity protocols.

Phased AI deployment strategy

“We didn’t just want to automate tasks, we wanted to reimagine how work gets done,” says Odity Mukherjee, who leads 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.”

Security requirements shaped the deployment approach. The agency handles proprietary pharmaceutical insights and needed trusted authentication systems and isolated project environments for client data protection.

Integration within existing Microsoft applications enabled faster adoption across Cactus Life Sciences teams.

Custom agents for workflow automation

The deployment included more than 30 custom automation agents designed to handle discrete tasks rather than replace entire job functions. Each agent targets specific repetitive processes within larger workflows.

The custom agents retrieve and structure information from scientific literature. Additional automation tools perform quality control functions including abbreviation verification, formatting standardisation and regulatory compliance checks.

These agents can extract and compare data across multiple complex documents. According to Cactus Life Sciences, structured data extraction is now 35% to 50% faster than previous workflows.

Project managers use the agents to summarise email threads and generate task lists automatically. Scientific writers process larger volumes of articles within shorter timeframes.

Internal adoption and training

Cactus Life Sciences created a Copilot Champions programme to identify internal advocates who could encourage experimentation with AI tools. 

A dedicated training team developed modular education programmes covering both capabilities and limitations of the AI tools. The company built a centralised repository to share prompts, agent concepts and implementation 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.”

Human oversight is maintained on top of all automated processes. Technology supports human professionals but does not operate independently and all automated outputs undergo review before deployment.