Key Takeaways

  • A business is ready for AI when it can name a task that costs time every week and has someone who owns the result.
  • You don't need perfect data or an IT department, but the documents for that one task should be in order.
  • Before the first pilot, settle which data may go into which tools.
  • Decide in advance how you'll measure success, or the result stays a matter of opinion.
  • Your answers show whether to start a pilot, lay the groundwork first or begin by understanding your workflows.

A business is ready for AI when it can name a task that costs time every week, the documents for that task are in reasonable order, it's clear which tools and data are allowed, and someone owns the result. You don't need perfect data or your own IT department. The ten questions below show where you stand and what to do first. Most of them can be answered in an hour.

What does AI readiness mean for a small business?

In large corporations, AI readiness often means data platforms, architecture and specialist teams. For an SME it's far more down to earth, and it has three parts. First, the work: a clearly defined task where AI can help. Second, the foundations: documents, systems and rules. Third, the people: time, ownership and a willingness to try something new. The ten questions cover all three. Answer them honestly, ideally together with one or two people who do the work every day.

Which ten questions show where you stand?

  1. Can you name a task that costs time every week?A good answer is a specific, recurring task with a rough idea of how often it comes up: answering tenant inquiries, putting quotes together, chasing missing receipts. “We should do more with AI” isn't a task yet. If you have no answer, ask your team for a week which work they'd most like to be rid of. Our page for when you don't know where to start helps narrow it down.
  2. Do you know how that task runs today?Who does which step, in which system, with which exceptions? If the answers differ within the team, that's already an important finding. Before you automate anything, map the process, as described in our five-step workflow audit.
  3. Can you find the documents the task needs?Templates, price lists, past cases and internal rules should be up to date and kept in one shared place, not scattered across personal inboxes. You don't need to tidy up the whole server. Sort out only what this one task needs.
  4. Do you know which data may go into which tool?According to the Federal Data Protection and Information Commissioner (FDPIC), the Federal Act on Data Protection is directly applicable to AI-supported data processing. Anyone bound by professional secrecy, such as a lawyer, has to apply stricter standards on top. If there's no clear rule yet, write a short AI usage policy. This is a general pointer, not legal advice.
  5. Can your main systems be connected?Inbox, file storage and industry software: do they run in the cloud or on your own server, and are there interfaces or at least exports? That decides how much work an integration will be. If it's unclear, start with a task that works from emails or documents, and settle the systems question in parallel.
  6. Is your team already using AI, and do you know how?Some people may well be using AI tools on their own already, perhaps with private accounts. That shows interest, but it becomes risky once customer data is involved. Ask openly and without blame: what do you use, for what, and what has worked? The answers show where rules and training need to start.
  7. Who would own the first project?You need someone from the business side with time and room to decide, who defines what a good result looks like and sees the pilot through. If everyone is responsible, nobody is. In small businesses this is often the owner, though sometimes the person who does the task every day is the better choice.
  8. Is leadership backing it with time and budget?A pilot needs a fixed budget, protected hours in the team and someone in leadership who decides at the end. What AI costs an SME explains which costs to expect.
  9. How will you know it worked?Before you start, decide what should improve and how you'll measure it: time per case, response time, errors. Without a baseline, every result is a matter of opinion. A simple approach is described in measuring time saved by AI.
  10. Do you want the solution built for you, or to learn to build it yourselves?Some businesses want a finished solution, others want their team to be able to use and develop AI themselves, and many want both. There's no wrong answer, but it sets the route. Build or train? shows how to weigh it up.

How do you read your answers?

Don't add up points. Look at where the gaps are, because that tells you the next step:

  • Mostly yes: you're ready for a pilot. Take the task from question 1 and follow a clear plan, such as our 30-day plan for where to start with AI.
  • Yes on the work (questions 1 to 3), no on rules and systems (4 to 6): lay the groundwork first. Approved tools, a short data rule and a training session can be in place within a few weeks.
  • No on question 1 or 2: start with the workflows, not the tools. A workflow audit shows where the time goes and where AI can help.
  • No on questions 7 to 9: settle ownership, budget and how you'll measure success before anyone builds anything. Otherwise you end up with a pilot nobody makes a decision about.

An example: a property management firm answers question 1 with “tenant inquiries by email”, knows the process well and has templates for the most common replies. What's unclear is whether tenant data may go into the assistant some of the team already use. So the next step is a short data rule and the choice of an approved tool, and no project is needed yet. After that, little stands in the way of a pilot. Tenant inquiries in property management shows what such a pilot looks like.

What doesn't AI readiness require?

  • Perfect data. It's enough for the documents behind the first task to be in order.
  • An IT department. A first pilot can often run on the systems you already have, and an external partner can take care of the technical side.
  • A big budget. A pilot with a clear scope shows whether more is worthwhile.
  • A fully worked-out AI strategy. The strategy gets better when it builds on what you learned in a first pilot.
  • A team that's enthusiastic to a person. Scepticism is normal and often justified. It tends to fade when people see a solution taking work off their plate.

You're ready for AI as soon as you know which task you want to change and who will answer for it.

What's a good first step after the check?

Take the task from question 1 and write on one page what happens today, what should improve, who is responsible and how you'll measure success. With that page, you can decide internally or talk to a partner. If you'd like to go through the ten questions with us, get in touch.

Frequently Asked Questions

How long does an AI readiness check take?

An hour with leadership and one or two people from the team is enough for the ten questions. If it turns out that a process needs mapping first, add a few days for observing and measuring. A thorough analysis of several processes is a project in its own right.

Do we have to clean up all our data first?

No. Tidying up without a goal costs a lot and achieves little. Put in order the documents your first task needs, such as current templates and the valid price list. Whatever works there, you can later apply to other areas.

Is our business too small for AI?

Size is rarely the problem. What matters is whether there are tasks that repeat and take time, and those exist in an office of five people too. Small businesses even have an advantage: decisions are made faster, and everyone sees straight away whether a solution helps.

Who should answer the questions?

Leadership together with people who do the work in question every day. Their answers often differ, especially on how a process runs today. Those differences are exactly what makes the exercise useful.

What if we answer no to most of the questions?

Then you know where to start. Begin with the first question you answered no to, because the order follows the logic of a project. The groundwork can often be laid in a few weeks, and after that the check already looks different.