Starting from the Work, Not the Question

The nuclear professionals I’ve met over 20+ years in the industry don’t open an AI tool because they’re curious about AI.

They open it because they have work to finish.

• Write a report
• Update a procedure
• Answer an inspector question
• Find a reference
• Understand what changed
• Get a relief request approved

Decades of search tools have conditioned people to believe the box wants a question. What they actually have is an outcome they’re responsible for delivering.

Think intent, not questions

I’ve found that our users get better results when they stop trying to compress their work into a search term.

They start by describing the job they’re doing. What's their role? What are they trying to create, whether it's a document or a decision? What's the context and who's going to read it?

This doesn’t require polish. It doesn’t require clever phrasing. It works best as a straightforward set of sequential instructions designed to get you to the outcome.

Treat the system like your coworker

A useful mental model: you’re the mentor to a capable coworker with access to the full company library.

I’ve mentored dozens of people and led large organizations, and I’ve never expected someone in a new situation to deliver a perfect answer with no context. Instead, good mentors show what good looks like, and they know there’s going to be some course-correction.

So treat the AI like that: explain the task, give the context, and be explicit about what output you need. When you get your first answer, use follow-ups the same way you would with that junior engineer:

• calling out what’s missing
• asking for another angle
• requesting more specificity
• tightening language or scope

The key habit is keeping your eyes on the prize, not judging the first draft.

When you end up with a great result through this coaching, switch roles and ask for advice on how to frame the task better next time. That turns a one-off interaction into a repeatable starting point.

To recap: the habits that make the difference

This matters now. I’m seeing a dwindling set of experienced nuclear professionals and intense pressure to get things done.

People who use AI to bridge this gap don’t do anything fancy. They do two things:

• They think in terms of intent, not questions.
• When in doubt, they start.

You can always follow up. You can always refine. What matters is getting the work moving in the direction you need—using the same judgment and standards you already apply everywhere else in nuclear work.

You don’t need to learn how to prompt.

You already know how to get work done.

If you want to see how this works in real life, reach out to me—theresa@everstar.ai.