Building a Responsible AI Framework for MOSAIC BC.

The challenge

MOSAIC BC is a not-for-profit organization offering settlement and employment services for newcomers to Canada. With a desire to be true to their mission, efficient with their resources, and empowering to their staff, leaders at MOSAIC knew AI tools could play a big role.

The question was how. How to get staff trained and equipped with guardrails that guide responsible, effective AI use across an organization with diverse programs, sensitive client data, and a strong values orientation. That's the question MOSAIC asked Sam to help explore.

Our approach

Start with what's actually happening

We started with a survey to get the lay of the land. It revealed that 93% of respondents were already using some version of AI in their regular work. Among the leadership team, that number was slightly lower at 87%. Staff identified broad benefits, improved efficiency, communication, and creativity, alongside natural concerns about safety, accuracy, and where the lines should be drawn.

Many staff members were already looking for training and organizational guidance. They wanted to use AI well, and they wanted MOSAIC's leadership to define what "well" meant.

Build a four-layer Responsible AI Framework

We co-developed the MOSAIC Responsible AI Framework, designed to ensure ethical, secure, and effective AI adoption across the organization. The framework is built on four interconnected layers:

• Principles, values-aligned commitments that guide every AI decision MOSAIC makes
• Policy, staff-facing guidelines that translate principles into specific dos and don'ts
• Oversight, governance structures and review processes that keep AI use accountable
• Processes, practical workflows for evaluating new AI projects, including a project assessment tool staff can use directly

The framework includes draft policies and ready-to-go templates. Practical, staff-facing guidelines that are easy to use day-to-day, allowing the organization to move forward with greater clarity and momentum.

What's next for MOSAIC

• Implementing AI solutions where the value is high and the risk is low. The framework gives MOSAIC clarity on which use cases to greenlight first.
• Equipping internal champions to lead. A small AI Interest Group is taking ownership of cross-organizational discussions, promoting safe and effective AI use, and identifying the next set of opportunities.

Why this approach worked

• Evidence before policy. The survey grounded every framework decision in what staff were actually doing, not what leadership assumed.
• Practical over theoretical. Templates and tools that staff can use directly, not abstract principles that need translation.
• Internal ownership from day one. The AI Interest Group means MOSAIC's framework keeps evolving with the work, not stuck at the version we shipped.