I hear the same complaint from a lot of leaders: “We tried AI and the results were generic.” Sometimes that's fair. More often, the tool did exactly what it was asked. The request just wasn't clear, because the work itself had never been clearly defined.
Think about what you'd tell a sharp new hire on their first day. You wouldn't say “write a proposal” and walk away. You'd explain who the customer is, what they care about, what a good proposal looks like here, and what we never promise. AI needs the same thing. Most organizations have never written that down for people either.
The real work comes first
Before AI can help with a task, somebody has to answer a few questions:
- What is this task actually for?
- What does a good result look like? Do we have examples?
- What information does it depend on, and where does that information live?
- Who checks the result, and what are they checking for?
If your team can't answer those, AI will expose it fast. That's not a reason to avoid it. It's one of the more useful things it can do.
Where I'd start
Pick one task your team does every week that takes real time and follows a rough pattern. Meeting summaries, first drafts of customer emails, research before a sales call, and training materials are all good candidates.
Write down how a good person does it today. Gather two or three examples of the work done well. Then build the AI step around that, and have the person who knows the work review what comes out.
Do that a few times and you end up with more than a few saved hours. You end up with processes that are clear enough for a machine to follow, which means they're clear enough for your next hire too.