Our full-stack, cross-discipline product team joined the life sciences company to design a system for managing country trial planning at scale. The platform combined three pillars: 

• AI models trained on historical enrollment data
• Constraint-optimization engine to balance speed and cost
• A web application that gave analysts direct control

Together, these elements turned country strategy design from a fragmented manual exercise into a structured, data-driven process.

1. Research and discovery

Before writing a single line of code, our team interviewed feasibility analysts, therapeutic leads, and medics, the people who lived the country trial strategy process every day. They walked us through how strategies were built and discussed pain points around gathering enrollment data, weighing regulatory requirements, and debating site capacity.

To untangle this complexity, we created a service blueprint that captured roles, the sequence of decisions, the terminology analysts used, and the points where delays crept in. We also framed what success would look like with clear goals, such as:

• Reducing time to create country strategies by weeks
• Providing recommendations that analysts can trust
• Having enough flexibility in the solution to handle edge cases 
• Demonstrating a clear impact on business ROI

This blueprint and the opportunities for innovation it yielded became the foundation for a high-value full-scale web application.

2. AI forecasting and constraint optimization

With the blueprint in place, we turned to intelligence. The pharma company’s data scientists brought historical trial data to the table with years of enrollment patterns across geographies. This helped train machine learning models capable of predicting how quickly patients could be enrolled in different countries.

Next, we paired the ML model with a constraint-based optimization engine. For the first time, analysts could ask the system real-world questions and get answers instantly, such as, What’s the fastest trial if we must include Japan? How do costs change if we cap patients per site at ten? What happens if we exclude a country due to political risk?

This turned strategy design from a static, linear process into an iterative scenario-based one, where analysts could explore, challenge, and validate, with AI as a partner rather than a black box.

3. Building an intelligent web application

Once the models and optimization engine were in place, we focused on delivery. Our team designed and built a cloud-hosted application on Heroku (running on AWS) that served as the front end for the system. The build process prioritized scalability, security, and a clean user experience, ensuring the application could integrate seamlessly into analysts’ workflows.

The new web application allowed the life sciences firm’s analysts to:

• Enter trial parameters and constraints
• Generate optimized strategies in minutes
• Visualize enrollment curves, projected costs, and site distributions by country
• Run “what-if” scenarios without starting over
• Override AI outputs with their own expertise

The application was shaped by clear design imperatives from discovery: don’t oversimplify, preserve statistical integrity, and make recommendations transparent. Analysts could see not only the outputs but the reasoning behind them. 

By codifying institutional knowledge into a repeatable system and surfacing it through an intuitive interface, the web app redefined how strategies were created. Analysts were freed to focus on the decisions that mattered most, while the system handled the complexity in the background.