Most companies start with AI the wrong way: they buy a tool, then look for a problem to give it. A few months later the tool is forgotten and the team says "AI does not work for us".
It works better the other way around. Start with your business, then choose the tool. These five questions take less than one hour to answer, and they will save you months.
1. Which task eats your week?
Do not start with "we need a chatbot". Start with a list of the tasks your team repeats every day: answering the same customer questions, copying orders from WhatsApp to Excel, writing the same quote again and again.
Pick one task that is repetitive, costs real hours, and follows clear rules. That is your first AI project. One task, not ten.
2. Is the data already written somewhere?
AI learns from examples. If you want it to answer customers, it needs your past answers. If you want it to read invoices, it needs your invoices.
Ask yourself: are these examples already written — in emails, messages, files, a database? If everything lives in one person's head, your first project is not AI. It is writing things down.
3. What happens when the AI is wrong?
AI makes mistakes, and it makes them with confidence. So the real question is not "is it perfect?" but "what does a mistake cost?".
Choose a first task where a human checks the result before it reaches a client: a draft reply that someone approves, a summary that someone reads. Keep AI away from final decisions about money, contracts, or health until you trust it.
4. Who owns it after launch?
An AI tool is not a machine you install and forget. Someone has to read its answers, correct it, and update its instructions when your prices or products change.
Give it one owner by name. No owner means that in three months nobody is using it.
5. How will you measure the win?
Before you start, write down one number: hours spent on the task each week, average time to answer a customer, number of errors per month.
Measure it again after one month. If the number did not move, stop or change the approach. If it moved, you have proof — and a good reason to automate the next task.
The one-page checklist
Copy this list and fill it in before you talk to any AI vendor:
- The task: one repetitive task, described in one sentence.
- The data: where the examples are stored today.
- The risk: what a mistake costs, and who checks the result.
- The owner: the name of the person responsible after launch.
- The number: what you measure today, and the target after one month.
If you can fill in these five lines, you are ready. If one line stays empty, you know what to fix first.
Need help with the answers?
At GenZ Apps we build custom applications and AI agents for companies, starting from the task, not from the tool. If you want a second opinion on your five answers, tell us about your project — we reply within 24 hours.