1. Is there a repeated task worth improving?

Start with work that is repeated every day and follows rules: answering the same questions, sorting requests, preparing quotes, writing reports. AI creates value where the task is frequent and the cost of delay or error is visible.

2. Is the information it needs available?

An assistant can only answer from what it is given. Check that your price lists, policies, product details and past answers exist in a form someone can maintain, even if it starts as a document.

3. Is the process written down?

If your team cannot describe how a request moves from arrival to completion, automation will copy the confusion. Sketch the steps first, including who decides at each point.

4. Who owns it?

Every AI system needs a person responsible for its answers, its limits and its review. Without an owner, quality slowly drifts.

5. How will you know it works?

Choose one or two measures before you start, such as time to first reply, number of requests handled without staff, or leads passed to sales with complete information.