Half of Greek Firms Say They Use AI. Few Use It Well.

Nearly half of Greek businesses now say they use AI. It's a good headline. It also hides the more useful question: is the AI doing any real work?

Here's what the numbers show, why so many firms stall, and what the rules actually ask of you. At the end there's a 10-question self-check you can do in two minutes.

What the numbers show

The headline comes from Unlocking Greece's AI Potential 2026, a survey by Strand Partners commissioned by Amazon Web Services, covering 1,000 business executives and 1,000 citizens. It found that 47% of Greek businesses use AI, up from 34% a year earlier.

Among those users, the study reports three levels:

If those shares apply as reported, advanced users are roughly 8% of all Greek firms (17% of 47%, our arithmetic).

Now the less flattering number. AWS sells cloud services, and "do you use AI?" is a generous question. Eurostat's official survey, which covers firms with 10 or more staff, puts the EU at 20% in 2025 and, as reported by Naftemporiki, Greece at around 9%, with smaller firms at about 8.5%.

The two figures measure different things, so neither one is simply wrong. Together they say the same thing: many people in Greek companies are trying AI, and far fewer companies have built it into how they work.

Why firms get stuck at the basic tools

Skills. In the AWS study, 48% of firms cite a shortage of AI and digital skills, and only 16% believe they have strong AI skills. A subscription doesn't fix that.

Data. AI can only help with what it can reach. If invoices live in one inbox, customer history in someone's head and prices in an old spreadsheet, every task starts with copying and pasting.

Trust. Owners are right to ask who checks the output and where client data goes. Without an answer, the sensible choice is to keep AI away from anything that matters.

Rules. Headlines about new AI laws make it easy to wait. As you'll see below, most small firms have less to fear, and one simple thing to do.

What better use looks like

These are illustrations of the kind of work we mean, not client stories.

• A small accounting office. Client emails with attached receipts are sorted into the right client folder, and missing documents are listed for a person to chase. The accountant still checks every figure.
• An online shop. Staff get a draft answer to "where is my order?" or "can I return this?" built from the shop's own orders and return policy. A person approves it before it goes out.
• A small logistics firm. Every Monday, one short report lists late deliveries and unusual costs from data the firm already has, so the manager starts with the exceptions.

Notice what they share: one recurring job, the firm's own data, and a person who stays responsible.

What the rules actually require

This is our plain reading of public texts, not legal advice.

The Greek AI-at-work bill mostly concerns platforms. The Labour Ministry's draft is open for comments until 19 November 2026. Press coverage makes it sound as if it covers every workplace, but the draft's scope article applies those provisions to digital labour platforms organising work in Greece. They would have to explain their automated systems to the people working through them and keep people overseeing those systems. If you run a delivery platform or supply its couriers, read it closely. It's a draft, so the text can still change.

AI literacy applies to you now, in a lighter form. Under Regulation (EU) 2026/1744, companies using AI must take measures to support the AI literacy of their staff. The earlier duty to ensure a sufficient level was softened, and the text says no specific level has to be guaranteed for any individual. Greece's Law 5321/2026 makes the Data Protection Authority the main supervisor, and literacy measures can be taken into account when sanctions are set. So keep a simple record of what you did.

AI in hiring and managing staff has more time, but not forever. Using AI to filter candidates, allocate tasks based on behaviour, or evaluate employees is generally high-risk under the AI Act. Those obligations now apply from 2 December 2027. They include telling workers and their representatives before using such a system at work. Data protection rules already apply today.

Funding: wait for the published call. The government has announced a €150 million scheme for AI in small businesses. We haven't seen a published call, so don't plan around it yet.

Three steps for a small business

1. List what's in use. Which AI tools, who uses them, and what company or customer information goes in. An hour and a spreadsheet will do.
2. Train on real tasks and write one page of rules. Use your own emails, documents and jobs. Write down what may go into AI, who checks the output, and what's off limits. Keep a note of who attended.
3. Automate one job and measure it. Pick something that repeats every week. Time it before, time it after, and only then decide on the next one.

Free self-check: where is your company?

Give yourself one point for each honest yes.

1. Can you name every AI tool used in your company?
2. Do you know what company or customer information goes into them?
3. Do you have written rules for using AI at work, even one page?
4. Has your team been trained on its own real tasks, not only general tips?
5. Do you have a record of that training?
6. Does someone check AI output before it reaches a customer?
7. Is AI connected to at least one tool you use every day, such as email, documents or your CRM?
8. Does at least one weekly job now run with little manual work?
9. Have you measured the time that job saves?
10. If you use AI in hiring or managing people, do you know what those tools do and what their suppliers promise?

0–3: Starting out. People may be experimenting, but the company has no shared approach. Begin with steps 1 and 2.

4–7: Stuck at the basics. The foundations are close. Pick one job and automate it properly.

8–10: Ahead of most. Look for the next job worth connecting, and keep your records up to date.

This is a quick guide of our own, not a legal assessment or a scientific score.

Want a second pair of eyes?

If you'd like someone to go through it with you, our AI Check-up is a 60-minute call and a one-page plan naming the three jobs in your business worth automating first. It costs €149, and that comes off the price of any build you book within 30 days.