Optimizing Admissions at EHN Canada with AI

Meet the Team

Hans Bisoo	Lynn Liu	Timmy Williams

Project Brief

As part of the final EHN Canada AI team, we were engaged by EHN Canada, a national leader in addictions and mental health treatment - including depression, anxiety, trauma and PTSD.

EHN’s mission-driven care spans several specialized programs, and they interact with over 10,000 inquiries per month through its Admissions Team alone.

Our challenge was to identify and optimize a bottleneck in the discovery dialogue process used by the Admissions Team. These high-volume, often time-sensitive conversations require accurate and empathetic responses, crafted from both clinical knowledge and the specific offerings and nuances of EHN’s programs. Given the scale and complexity, the team needed a more efficient way to retrieve internal knowledge during live interactions.

We designed and developed an LLM Platform to augment internal admissions workflows. At its core is a real-time chatbot, trained on EHN’s internal knowledge bases and optimized for internal use. The system includes a simple interface for updating the knowledge base, allowing non-technical staff to keep content current without relying on developers.

To ensure reliability and alignment with internal needs, we integrated DeepEval as part of our continuous evaluation pipeline. This helped us quantify metrics (hallucination, bias, accuracy), test retrieval performance, and select default models that were both efficient and accurate. Our platform leverages open-source tools and locally-hosted models, reflecting our commitment to privacy, transparency, and long-term maintainability.

The result is a scalable, secure, and customizable AI support layer for EHN’s Admissions Team—reducing information retrieval friction, increasing consistency across staff responses, and freeing up the admissions staff to focus on high-empathy, high-impact conversations.