CanITCon2026 · September 11, 2026 · Calgary, CA

Nobody Buys “AI Readiness.” Everybody Needs an AI Operating Model.

For two years, the MSP playbook for AI has been the readiness assessment: a one-time $5,000 to $15,000 project that scans a tenant, produces a deck, and then goes quiet. It sells once. It sounds like 2005-era consulting. And it leaves the client exactly where the MSP found them, aware they have a problem but no closer to an outcome. The uncomfortable truth is that nobody actually wants to buy AI readiness. Readiness is a status, not a result. The real problem is not a tools problem. MSPs are surrounded by tools: Copilot, Claude, ChatGPT, an entire Agent Store. The problem is an operating-model problem: how to sell, deliver, govern, and measure AI as a managed service across an entire client base, the same way every time. Building custom agents for every client does not scale. The average MSP believes it can meaningfully serve maybe 20 percent of its clients with AI in a given year, which leaves 80 percent of the book untouched while shadow AI spreads inside those same companies unmanaged and ungoverned. This session makes the case for the shift from a project to a practice. We will walk through what an AI operating model actually looks like inside an MSP: a repeatable motion that discovers demand and shadow AI, plans the managed adoption of generative AI, then runs deployment, governance, and observability on an ongoing basis. We will cover the recurring rhythm that keeps it alive, a virtual Chief AI Officer accountable for the roadmap, a monthly working session where decisions get made and usage gets reviewed, and a quarterly segment that proves value back to leadership. The through-line is measurement: training completion as the leading indicator, actual observed usage as the truth, and adoption tracked against a baseline rather than assumed. Attendees will leave able to answer the question every client is really asking, which is not "are we ready" but "what do we do now, and how do we know it worked." The economics follow the model. A one-time assessment is a $10,000 event. An AI operating model delivered as a managed service is recurring revenue that gets measured, managed, and renewed. Readiness gets you in the room. The operating model is what you actually sell.