AI-agent-first development: Transforming developers into AI coders at MedIT

Primary speaker:

Maurice Tam, Lead AI Developer & Systems Architect, MedIT - Temerty Faculty of Medicine

Additional speakers:

Philip Chung, Associate Director, Applications and Data Architecture, MedIT

Sam Xu, Developer, MedIT

Description: 

MedIT is exploring how artificial intelligence is transforming the way software is designed, developed and delivered. This session presents a real-world case from the MedIT Development Team at the University of Toronto’s Faculty of Medicine, illustrating how AI tools are reshaping developer roles, development workflows and software quality outcomes.

Since 2025, MedIT's Application Engineering and Systems Architecture team has integrated AI (primarily Claude) into the entire software development lifecycle: planning, analysis, coding, testing and deployment. This shift enabled a transformation from traditional vibe coding toward context-driven development, where developers collaborate with AI to generate structured solutions based on system architecture, domain knowledge and policy requirements.

As a result, our development model is evolving from Software Developer to AI Coder, where developers focus more on domain expertise, system context and AI orchestration while AI assists with implementation, testing and documentation. This AI-agent-first development approach has significantly improved productivity and increased software quality by an estimated threefold.

We showcase a Medicine faculty-level AI-powered system prototype: a multi-agent AI workflow for Clinical Appointment management. The system demonstrates how specialized AI agents collaborate to support complex academic workflows, including:

  • Credential validation
  • Policy and guideline retrieval
  • CV information extraction
  • AI-assisted academic rank recommendation

A short demonstration of the proof-of-concept system will illustrate how AI agents can streamline administrative and academic processes in healthcare education environments.

The session concludes with a Q&A. Attendees will gain practical insights into adopting AI-assisted development practices, designing AI-agent workflows and preparing development teams for the emerging era of AI-driven software engineering.

Go to Top