The leave-behind principle

The best measure of an AI engagement isn’t what you ship. It’s what happens after you leave.

Most delivery models — in any industry — quietly optimize for the opposite. Dependency is recurring revenue. Capability transfer shortens the engagement.

That’s why the design choice in AWS’s Forward Deployed Engineering launch is worth reading closely. Customers exit with agentic systems running in their own environment — and with the skills, workflows, and patterns to keep building on their own. Knowledge graphs. Runbooks. Trained internal champions. As Francessca Vasquez put it, customers “gain lasting AI skills, workflows, and patterns they can use to innovate independently.”

I’ve watched this play out with startups for years. Founders remember which partners made their team stronger and which ones made themselves indispensable. The first group gets the next call. The second gets audited.

Enterprises are no different — they’re just slower to say it out loud.

If you’re evaluating any AI delivery partner this year, ask one question before price or timeline: “What does my team know how to do the day you leave that it didn’t know the day you arrived?”

The answer tells you everything about the model you’re buying.