2 Questions That Reveal Your Perfect AI Use Cases

Most teams don’t have an “AI strategy” problem. They have a clarity problem. When you stop talking about models and start talking about work, the best use cases basically reveal themselves.

When we onboard a company at Notis, we run a simple interview with every role. No jargon. No hype. Just two questions that surface the low-hanging fruit fast.

The two-question interview (and why it works)

Here’s the trick: people already know what’s broken in their week. They don’t need a workshop. They need permission to say it out loud.

So we ask everyone—from finance to sales to ops—the same two questions. The answers cluster immediately, and those clusters are your automation backlog.

Question 1: What do you wish you did more?

Ask this to anyone on your team and you’ll get an answer in seconds. It’s usually the work they’re proud of, the work that moves the needle, or the work their manager keeps nudging them toward—but it keeps losing to the urgent.

Common examples: posting consistently on social, reviving the blog, shipping a monthly newsletter, following up with warm leads, documenting processes, actually looking at user feedback instead of “saving it for later”.

This question matters because it tells you where leverage is. If AI can help someone do more of a high-impact activity without increasing their hours, you don’t just save time—you increase throughput.

Question 2: What do you wish you did less?

This one is even faster. People instantly point at the parts of their job that feel like friction: repetitive admin, formatting, searching, copying, chasing, rewriting the same thing for the tenth time.

The examples I hear all the time: “I track expenses every month and it takes me four hours.” “Before a client meeting, I spend ages preparing—finding the right deck, pulling case studies, remembering what was said last time.” That’s your automation gold.

If something is predictable and time-consuming, it’s usually a great candidate for AI assistance: extract data, draft first versions, summarize, pre-fill, pre-structure, and hand you the 80% so you can do the 20% that requires taste and judgment.

How to turn two answers into an AI roadmap

Once you have the answers, the goal isn’t “automate everything.” The goal is simple: save each person roughly a day per week on the stuff they hate, so they can reinvest that time in the stuff they know they should be doing.

Here’s the practical way to do it.

1) Collect answers per role (15 minutes each)

Run quick 1:1s with the two questions. Capture examples in the person’s words. The wording matters because it tells you what “done” looks like.

2) Group the answers into repeatable workflows

You’ll see patterns: meeting prep, expense tracking, follow-ups, content repurposing, reporting, documentation. Those patterns are where automation actually sticks.

3) Start with “assist”, not “replace”

The best first wins are copilots: summarizing, drafting, extracting, pre-filling, creating a first version. Humans keep the final decision and the taste. That’s how you get adoption without breaking trust.

4) Measure impact in hours saved and output shipped

If you can say “we saved sales 3 hours/week of meeting prep” or “marketing ships 4 posts/week instead of 1,” you’re not experimenting anymore—you’re compounding.

If you only take one thing from this: don’t start by asking “where can we use AI?” Start by asking your team what they want more of, and what they want less of. The use cases will practically write themselves.