AI now comes up in almost every management meeting. Some companies have run successful pilots that saved real working hours; others spent budget on tools and subscriptions nobody used after the first month. The difference rarely lies in the tool. It lies in the questions asked before starting.
Question 1: What exact problem are we solving?
Start from the work, not the tool. "We want to use AI" is not a problem. "The customer service team spends half its day answering repeated questions" or "preparing a quotation takes two days" are problems you can measure. Write the problem in one sentence, name who suffers from it, and estimate how much time or money it costs today.
If you cannot describe the problem that precisely, the project is not ready yet.
Question 2: Do we have the data and a clear process?
AI does not fix a chaotic process; it speeds it up as it is. If every employee prepares the report differently, or customer data is spread across scattered files, the first step is to standardize the process and organize the data. Often the team discovers during this step that half the problem is solved before any tool is introduced.
Ask, too: is the data sensitive, and what is the company's policy on uploading customer data to external tools? That is a governance question to settle before the pilot, not after.
Question 3: How will we know it worked?
Before starting, choose one success measure: working hours saved per week, customer response time, or errors in documents. Then set a short pilot of four to eight weeks with a small team. If the improvement appears, expand. If it does not, stop without having committed to a long contract or a company-wide change.
Apply it: a one-page project card
- Problem: one measurable sentence.
- Who suffers from it: the departments and roles involved.
- Current state: how much time or money it costs today.
- Process and data: is the process standardized? Where is the data? Is it sensitive?
- Success measure: one number measured before and after.
- Pilot length and team: a set number of weeks and a small team with a named owner.
Where companies usually start
The safest starting points are repetitive writing and analysis tasks: drafting correspondence and proposals, summarizing meetings and reports, first replies to recurring customer questions, and analyzing spreadsheets. Time saved is easy to measure, and a person stays responsible for the final review.
A company that starts small and measures carefully builds real expertise inside its team within months, and that matters more than any tool.