Paul Boag's central insight, that we should treat AI like a skilled intern with no real-world experience, is the most practical mental model we have encountered for working with these tools. It strips away the hype and gets straight to what matters: AI can be eager, fast, and surprisingly capable, but it has no judgment, no context, and no instinct for when it is wrong. The mistake most of us make is expecting it to behave like a senior colleague or, worse, like magic. Boag's hard-won lesson is that the path to productivity runs through honest, humble delegation.
What does this mean for your team? It means you need to stop asking AI to do your job for you and start asking it to help you do your job better. An intern does not walk in and redesign your entire workflow. You give them a clear task, you check their work, and you accept that the first few tries will need correction. Boag's experience with wasted hours and terrible prompts is not a failure of AI, it is a failure of expectation. The practical shift is simple: treat your prompts like instructions to a junior. Be specific, provide examples, and anticipate that the output will require editing. The time you save is the time you would have spent on repetitive groundwork, not the time you would spend on strategic thinking.
This is where the model becomes truly useful. An intern can draft a user research summary, generate design alternatives, or produce a first pass at content. They cannot decide which research insight matters most, judge whether a design aligns with your brand, or know when a sentence carries the wrong tone. That is your job, and it always has been. Boag's real discovery is that AI does not replace expertise, it amplifies it. The teams that succeed are the ones that treat the AI as a force multiplier for their own skills, not as a shortcut around them. They review, refine, and direct, rather than hand off and hope.
The concrete takeaway is this: assign your AI a specific role on every project, with a clear deliverable and a clear review process. If you cannot describe what success looks like in two sentences, you are not ready to prompt. If you accept the first output without reading it, you are not delegating, you are abdicating. Boag's approach works because it is honest about the division of labor. Treat AI like an eager intern, and you will get more done. Treat it like a wizard, and you will keep chasing magic that does not exist.