Wave Agency

AI and Technology

Is Your Business Ready for AI? Five Honest Questions

Five questions about your work, your data, your process and your people to answer before you spend a dollar on AI tools.

A software rep gives a great demo. The tool reads an email, drafts a reply, updates the customer record and books the follow-up, all in about a minute. Everyone in the room nods. Three months later the subscription is still running and nobody uses it.

The tool was probably fine. What was missing was everything around it: a clear job for it to do, information it could actually read, a process that stayed the same from week to week, and someone whose job it was to check the work. Those things have nothing to do with which AI product you pick, and you can find out whether you have them in an afternoon, before you sign anything.

Here are the five questions we would ask first. Answer them honestly, in writing, with whoever actually does the work in the room.

1. Can you name the work, and what it costs today?

“We should be using AI” is not a starting point. “Someone spends most of Monday morning copying web inquiries into the CRM” is.

Pick the repeatable work that eats your team’s week: answering the same questions, entering the same data, writing the same kind of quote, chasing the same unpaid invoices. For each one, write down roughly how often it happens, how long it takes and who does it. Rough is fine. This is a baseline, not a spreadsheet model.

Without one, you will never know whether AI helped, only that you are paying for it.

A useful test: if you could hand this task to a capable new hire with a one-page instruction sheet, it is a good candidate. If the instructions would be “use your judgment,” it probably is not, at least not yet.

2. Is the information in one place a system could read?

AI works with what it can see. If your prices live in one person’s head, your customer history is split between an inbox and a paper folder, and the current version of the service menu is whichever PDF was emailed last, no tool will fix that for you. It will just be confidently wrong faster.

For each task on your list, ask where the information it needs lives. Is it written down, current, and in a system with a login rather than on personal devices? Most AI projects that stall do so here, long before the technology itself becomes the problem.

If this question makes you wince, start with the data, not the AI. We wrote a plain guide on where to start with scattered business data, worth doing whether or not you ever automate anything.

3. Does the process hold still long enough?

An AI tool follows the process you give it. If that process changes every month, because pricing is still being worked out, because a new manager is reorganizing the team, because nobody agrees on how refunds work, the tool will be following last month’s rules.

Look at each task and ask: has the way we do this changed in the last three months? Will it change in the next three? If the answer is yes, settle the process first. Write down how it works now, agree on it, and run it by hand for a few weeks. Then automate it.

That is not a reason to wait forever, just a reason to make a stable task your first one.

4. What must it never touch?

Every business has work where a wrong answer is expensive: patient or client records, payment details, contracts, anything regulated, anything that goes to a customer without a second look. Decide what those are before any tool is chosen, not after something goes wrong.

Make two short lists. The first is information the AI should never see at all. The second is actions it should never take without a person approving them, such as sending a message to a customer, issuing a refund, changing a price or deleting a record. Our pieces on what an AI agent should never do without asking and giving an agent the access it needs and no more go deeper on both.

In a regulated field, this question may decide which tools you can use at all. Better to know before the demo.

5. Who owns the output?

This is the one most businesses skip, and the one that most often decides whether the project survives.

Somebody has to read what the AI produces, at least at first, notice when it gets something wrong, fix the instruction and decide whether it is actually saving time. If that person is “everyone,” it is no one, and the tool will fade out the way the one in the opening did.

Name a person for each task. Make sure they have time for it, not just the title. And make sure the people whose day will change are part of the conversation early, because they know where the exceptions are. If you are thinking about a first agent, our guide to your first AI employee walks through how to pick the job and the owner together.

Score yourself

For each task you listed, answer yes or no:

  • We know roughly how often it happens and how long it takes.
  • The information it needs is written down, current and in one system.
  • The process has been stable for at least three months.
  • We have written down what it must never see or do.
  • One named person will own the results and has time to do it.

Five yeses is a strong first project. Three or four means you know exactly what to fix before you start. Fewer than that is not a failure. It is a list of work worth doing anyway, because every item on it makes the business easier to run with or without AI.

The best first project is usually the most boring one on your list: frequent, stable, low risk and owned. It is rarely the most impressive demo.

How we help

Our AI Readiness Assessment is a short, fixed-scope engagement that looks at exactly these things: where the hours actually go, whether a system could see the data, what AI must never touch, and who works differently afterward. You leave with a ranked list of opportunities, a sequence for what to do first and next, and guardrails in writing. Sometimes the honest finding is that you are not ready yet, and we would rather tell you that than sell you a build that was going to stall.

Want a second opinion on your list? Book a call.

← Back to The Journal

Work With Wave

Put a good idea to work.

If something here is useful, let's talk about what it would look like for your business. A short call with a senior member of the team.

Book a Call