Primary speaker:
Bahman Yaghoobi, Full Stack Developer, Education Commons, OISE
Additional speakers:
Chris Klammer, Technology Adoption Consultant
Description:
Small, decentralized IT teams across U of T often struggle to maintain documentation and automated testing due to limited capacity and large application portfolios. At Education Commons (OISE), we adopted agentic AI to generate end‑user documentation, technical documentation, and automated tests at scale. This session shows how AI can help teams meet professional standards sustainably and responsibly.
Since 2017, I have worked with multiple IT teams across U of T and have consistently seen the same challenge: small, decentralized teams struggle to maintain high‑quality documentation and automated test coverage while managing demanding operational workloads. At OISE’s Education Commons, our development team maintains a large portfolio with limited staffing, leaving little capacity for these essential practices.
Over the past year, we introduced agentic AI to address long‑standing gaps that traditional workflows could never meet. By developing effective prompting strategies, we were able to generate automated feature and unit tests, produce structured end‑user documentation, and create technical documentation for complex systems. This approach improved quality, sustainability, onboarding readiness and long‑term maintainability.
This session will demonstrate practical examples, effective prompting techniques, benefits, risks and lessons learned — offering a roadmap for how decentralized teams across U of T can leverage AI responsibly to strengthen engineering practices without additional resources.
