The news
What is AI for business and what do the official figures say?
Short answer
AI for business means using artificial intelligence inside business processes: analysing documents, generating text, forecasting demand or automating tasks with agents. In 2025, 19.95% of EU enterprises with 10 or more employees used it, according to Eurostat, ranging from 17% of small firms to 55.03% of large ones.
The figure comes from Eurostat’s Use of artificial intelligence in enterprises, based on the EU survey on ICT usage and e-commerce in enterprises that national statistics offices carried out in early 2025 (data extracted in December 2025). Use rose by 6.47 percentage points in a single year.
The gap between countries is wide. Denmark (42.03%), Finland (37.82%) and Sweden (35.04%) lead, while Romania (5.21%), Poland (8.36%) and Bulgaria (8.55%) trail. Spain is a useful close-up because its national statistics office, INE, publishes more detail than the EU tables.
According to the INE survey on ICT usage and e-commerce in enterprises, published on 22 October 2025, 21.1% of Spanish companies with 10 or more employees used AI in the first quarter of 2025, 8.7 points more than a year earlier. Among those with fewer than 10 employees, use rose 5.9 points to 13.4%. Services lead (25.7%), followed by industry (17.5%) and construction (11.4%). As of 12 October 2026 this is the latest edition published (in Spanish): the 2025–Q1 2026 edition is not out yet.
One caveat changes the reading. Banco de España surveyed around 6,300 firms in November 2024: 19.9% used some AI system, but 60% of them did so on an experimental or pilot basis, 34% moderately and only 6% intensively. Using AI is not the same as having it built into a process.
- 2 Feb 2025
Article 4 of the EU AI Act starts to apply
Organisations that use AI systems must take AI literacy measures for their staff.
- 7 May 2025
Banco de España publishes its analysis of AI in Spanish firms
19.9% of the firms surveyed use AI, but 60% of them are at the experimental or pilot stage.
- 22 Oct 2025
INE publishes the ICT survey of Spanish enterprises
21.1% of companies with 10 or more employees use AI in the first quarter of 2025, 8.7 points more than a year earlier.
- Dec 2025
Eurostat publishes the 2025 European data
19.95% of EU enterprises use AI, up 6.47 points; Spain stands at 20.27% on the same basis.
- 28 Jan 2026
The UK expands free AI training for every adult
The government’s AI Skills Hub offers free AI foundations courses, with a goal of 10 million workers trained by 2030.
- 24 Jul 2026
The EU Digital Omnibus on AI reframes AI literacy
Regulation (EU) 2026/1744 keeps the duty to take measures without requiring a specific level. National authorities supervise it from August 2026.
EU vs Spain
How many companies use AI across the EU, and how does Spain compare?
In 2025, 19.95% of EU enterprises with 10 or more employees used at least one AI technology, according to the Eurostat database. Spain stood at 20.27%. A year earlier it was below the EU average (11.31% against 13.48%); it almost doubled its share in twelve months to end up slightly above.
AI adoption by company size: the EU-27 and Spain
The picture by size is almost identical on both sides. The real gap is not between Spain and Europe but between large and small companies, and it repeats everywhere.
| Criterion | EU-27 | Spain | Spain vs EU |
|---|---|---|---|
| 10 or more employees | 19.95% | 20.27% | +0.32 points |
| Small (10 to 49) | 17.00% | 17.17% | +0.17 points |
| Medium-sized (50 to 249) | 30.36% | 30.10% | −0.26 points |
| Large (250 or more) | 55.03% | 57.46% | +2.43 points |
| 10 or more, a year earlier (2024) | 13.48% | 11.31% | −2.17 points |
Across the EU the spread is the striking part, from more than four in ten companies in Denmark to about one in twenty in Romania:
42.03%
of enterprises in Denmark use AI, the highest share in the EU
Eurostat, 2025
19.95%
of EU enterprises with 10 or more employees use AI
Eurostat, 2025
5.21%
of enterprises in Romania use AI, the lowest share in the EU
Eurostat, 2025
60%
of Spanish firms using AI are still experimenting or piloting
Banco de España, 7 May 2025
With INE’s Spanish detail, which goes down to micro businesses, the ladder runs from 13.4% of companies with fewer than 10 employees to 21.1% of those with 10 or more and 58.2% of those with 250 or more.
Why does INE say 21.1% and Eurostat 20.27%?
Real uses
What do companies use AI for, and which AI solutions for business do they buy?
Across the EU, companies that use AI apply it above all to marketing or sales (34.70%) and to business administration or management (31.05%), according to Eurostat. Logistics is the least common use (6.08%). Size changes the mix: 47.51% of large AI users apply it to ICT security, against 14.51% of small ones.
In Spain the order is reversed at the top: administration and management (29.75%) just ahead of marketing or sales (29.61%), according to INE’s artificial intelligence table (in Spanish). The percentages are calculated on companies that already use AI, and each one can have several uses.
| Criterion | Total | 10 to 49 | 50 to 249 | 250 or more |
|---|---|---|---|---|
| Administration and management | 29.75% | 25.54% | 36.76% | 45.16% |
| Marketing or sales | 29.61% | 27.11% | 33.53% | 39.36% |
| Production or service delivery | 22.39% | 19.25% | 26.08% | 37.76% |
| R&D or innovation | 22.25% | 19.62% | 26.27% | 32.79% |
| Accounting, controlling or finance | 20.52% | 18.84% | 22.93% | 27.69% |
| ICT security | 18.41% | 12.56% | 27.07% | 42.61% |
| Logistics | 7.53% | 6.38% | 7.23% | 17.30% |
Generative AI for business: which technologies companies use
Measured across all EU enterprises, the most widespread AI technologies in 2025 were text mining (11.75%), generating pictures, video or sound (9.55%) and natural language generation or speech synthesis (8.76%). Among large enterprises, text mining reaches 35.04% and natural language generation 31.68%.
INE gives the Spanish breakdown among companies with 10 or more employees that already use AI:
- Generating images, video or audio: 53.87%.
- Analysing written language, such as classifying documents or extracting data from text: 49.55%.
- Generating written or spoken language, or code: 43.80%.
- Converting speech into text: 36.04%.
- Automating workflows or supporting decision-making, the core of AI for business automation: 33.41%.
- Machine learning for data analysis: 29.36%.
Buy or build: are the best AI tools for business off the shelf?
Half of Spanish AI users (51.58%) buy ready-to-use commercial AI software. 26.86% turn to external providers who develop or adapt it, and 17.33% build it with their own staff. In other words, most of the AI tools for business in use today are general-purpose products.
My take: there is no single best AI for business. Subscriptions like the one I analyse in what ChatGPT Business includes and where it falls short handle general tasks such as drafting, summarising or searching. But the value of a specific process (getting paid sooner, never running out of stock, answering customers better) lies in your data and your systems. That is where AI solutions for business that connect an agent to your ERP or CRM make the difference.
Six in ten are still testing
AI for business does not start with choosing a tool: it starts with choosing a process that is costing you money today.
Barriers
What holds companies back from using AI?
Lack of expertise, across the board. Of EU enterprises that do not use AI, 14.21% had considered it (12.65% of small, 22.26% of medium-sized and 36.54% of large ones). Among those, 70.89% cite a lack of relevant expertise, 52.52% a lack of clarity about the legal consequences and 48.83% concerns about data protection and privacy. Only 20.68% say AI is not useful to them.
In Spain the knowledge gap is wider still: 81.29% of companies that considered AI but do not use it say they lack the necessary skills in-house, around ten points more than the EU figure. Privacy (53.55%) and unclear legal consequences (52.93%) follow.
| Criterion | Total | 10 to 49 | 50 to 249 | 250 or more |
|---|---|---|---|---|
| Lack of expertise in the company | 81.29% | 83.21% | 75.68% | 74.18% |
| Data protection and privacy | 53.55% | 54.19% | 50.42% | 58.28% |
| Unclear legal consequences | 52.93% | 53.98% | 49.91% | 48.64% |
| Costs too high | 48.28% | 48.68% | 47.92% | 42.10% |
| Availability or quality of data | 47.86% | 47.84% | 47.32% | 51.38% |
| Incompatible with existing systems | 34.05% | 33.96% | 34.80% | 31.43% |
| Ethical considerations | 24.12% | 23.95% | 24.81% | 23.54% |
| AI is not useful to the company | 14.16% | 14.92% | 12.71% | 7.02% |
My take: hardly anyone doubts that AI can be useful; what is missing is knowing how to use it and with what safeguards. Two of the top three barriers are legal, and both can be solved by design: data hosted in the EU, minimum permissions and a person who approves anything important. I explain it step by step in my guide to the EU AI Act and GDPR for SMEs.
Large companies trip over their data
Duties and support
Is AI training mandatory, and where can EU and UK businesses find support?
In the EU, yes: since 2 February 2025, Article 4 of Regulation (EU) 2024/1689, the AI Act, has required organisations that use AI systems to take AI literacy measures for their staff. For support, the most useful starting points are the European Digital Innovation Hubs in the EU and the free AI Skills Hub courses in the UK.
AI literacy: what the AI Act requires
Regulation (EU) 2026/1744 (Digital Omnibus on AI), dated 8 July 2026 and published in the Official Journal on 24 July, reworded Article 4. The duty remains: anyone who uses AI must take literacy measures that reflect their staff’s knowledge and the context of use. What changes is that a specific level no longer has to be guaranteed.
The European Commission clarifies three useful points: no certificate is needed, an internal record of the training is enough, and national authorities supervise this article from August 2026. I cover it in more detail, with the other key dates, in the AI Act guide for SMEs.
If you are based in the UK, the AI Act is EU law, but Article 2 also covers providers and deployers outside the EU when the output of their AI system is used in the EU. If you sell into the EU, check where you stand.
If you are looking at AI training for your team, look for three things:
- Role-based training: leadership (risks and use cases), the teams using the tools (which data must never go in) and whoever supervises an agent (how to review it and when to stop it).
- Practice with your company’s real tools and processes, not generic examples.
- A record of who was trained, when and on what, which you can use as evidence.
Where EU and UK businesses can look for support in 2026
In the EU, the European Digital Innovation Hubs (EDIHs) are one-stop shops co-funded by the Digital Europe Programme and national or regional sources. They offer the chance to test before you invest, training and skills development, and financing advice. The network renewed in 2025 with a reinforced AI focus, and you can find your nearest hub in the EDIH catalogue.
In the UK, the government announced on 28 January 2026 that every adult can take free AI foundations courses through its AI Skills Hub. The programme aims to give 10 million workers key AI skills by 2030, including at least 2 million employees of SMEs.
In Spain, as of 12 October 2026 I have found no new open calls for Kit Digital or Kit Consulting: two orders published in the Spanish Official Gazette on 28 January 2026 (TDF/38/2026 and TDF/39/2026) let leftover funds be reused in future calls, without opening any. For training, the steady route there is the annual Fundae training credit, applied as a rebate on social security contributions.
Be wary of “subsidised AI”
First step
Where should you start with AI for small business without wasting money?
Start with a process, not a tool: choose a repetitive task that costs you hours or money today, measure how it is done, map it and put a first custom AI agent only in the step that needs judgement. That tackles both problems the data shows at once: six in ten companies with AI still stuck in testing, and the lack of expertise.
It is what I do in my AI consulting for businesses with the MAPA method to map and automate processes with AI. It has four phases:
- Map: draw the process as it works today, exceptions included. Nothing is automated without an approved map.
- Analyse: decide, step by step, what to automate, what AI supports and what stays with a person.
- Pilot: build with real cases and run it in parallel with the team before switching it on.
- Audit: review results, exceptions and process changes every month.
If you run a micro business, you do not need new AI software for small business on day one: start by automating one step with the tools you already pay for.
AI agents for business: why a custom one rather than another subscription
An AI agent is not a chatbot: it receives a goal, queries your systems, prioritises and proposes the next action. If you are unsure what you need, the guide AI agent vs chatbot vs automation has a six-question test, and in what agentic AI is I explain what these systems can already do.
The AI agents for business that work best at first are modest: they read data, organise the work and explain why. That is how I design custom AI agents: with data in the EU, access only to what each task needs and a person who approves every important action.
Choose your area and size to see a sensible first use case, what to map before touching anything and how to measure whether it works:
If you prefer to do it yourself, the free resources include the MAPA template to draw your process and the checklist “Can AI read your website?”.
For your business
What does it mean for your business?
It means AI for business is no longer a bet on the cutting edge; it has become a matter of execution. This is how the work changes in each area:
LeadershipHigh impact
With 55.03% of large EU enterprises using AI and six in ten Spanish users still testing, the advantage is not in “having AI” but in getting a process into production. Choose one or two processes with an owner and a metric, and ask for results against a baseline, not demos.
MarketingMedium impact
Marketing or sales is the top use of AI in the EU (34.70% of companies that use it) and the second in Spain (29.61%). Generating content no longer sets you apart; being found when your customers ask AI does. I explain how in what GEO is and how it differs from SEO and AEO and in the Search & AI Visibility service.
Sales and customer serviceHigh impact
Sorting queries, prioritising opportunities and preparing replies is where a first agent pays off soonest, because there is volume and judgement. If you serve customers through messaging apps, check each platform’s rules for AI chatbots and always tell customers they are talking to an AI.
IT and dataHigh impact
In Spain, 47.86% of companies that considered AI and do not use it cite data problems, and 34.05% incompatibility with their systems. Your role: accessible data, minimum permissions and an inventory of tools. Standards such as the Model Context Protocol (MCP) make it easier to connect agents to your systems.
If you would like help deciding where to start, I am José Galán, a Search & AI Visibility consultant and AI agent designer: I work on one process at a time, with data you can check.
Action plan
What to do now with AI in your business
Do five things, in this order: two this week and three next quarter. None of them requires buying anything yet.
This week: take stock of the AI already in use
Ask each team which AI tools they use and with what data; there are probably more than you think. Set three basic rules: which data never goes in, who is responsible for each tool and which accounts belong to the company.
This week: start AI literacy
Run a role-based session and keep a record of who attended and what was covered. If you operate in the EU, that meets Article 4 of the AI Act, and it tackles the number one barrier everywhere: lack of expertise. Check the free or subsidised options available where you are, such as the UK AI Skills Hub or the Fundae credit in Spain.
This quarter: choose a process and measure the baseline
Use the picker above to choose a first case. Note how long it takes today, how often it happens and how many errors there are. Without a baseline you will not know whether AI works.
This quarter: map it and run a pilot
Map the process with the MAPA method and test a first agent in parallel with the team, with read-only access: it proposes and a person approves. If you would rather not do it alone, tell me about your case.
This quarter: compare and decide
After a few weeks of piloting, compare against the baseline. If the agent is consistently right, widen its scope; if not, adjust it or drop it. Deciding to stop also saves money.
Key points
- 19.95% of EU enterprises with 10 or more employees used AI in 2025, according to Eurostat; Spain, at 20.27%, is on the average.
- The gap is about size: from 17% of small EU firms to 55.03% of large ones, and from 13.4% of Spanish micro businesses to 58.2% of those with 250 or more employees.
- Six in ten Spanish firms using AI are still testing, according to Banco de España.
- The main barrier is lack of expertise (70.89% in the EU, 81.29% in Spain), not price or usefulness.
- AI literacy has been mandatory in the EU since 2025; EDIHs in the EU and the AI Skills Hub in the UK are good places to look for support.
- Start with a measurable process, map it and use an agent only where judgement is needed.
FAQ
Frequently asked questions about AI for business
How many companies use AI for business in the EU?
According to Eurostat, 19.95% of EU enterprises with 10 or more employees used artificial intelligence in 2025, up from 13.48% in 2024. By size, the share ranges from 17% of small firms to 30.36% of medium-sized and 55.03% of large ones. Denmark leads with 42.03% and Romania trails with 5.21%. Spain, at 20.27%, sits on the EU average.
What do companies use artificial intelligence for?
Among EU enterprises that use artificial intelligence, the most common purposes are marketing or sales (34.70%) and business administration or management (31.05%), according to Eurostat; logistics is the least common (6.08%). In Spain, INE puts administration and management first (29.75%), just ahead of marketing or sales (29.61%). Text mining and generating images, video, sound or text are the most widespread technologies.
What are AI agents for business and how do they differ from ChatGPT?
AI agents for business are systems that receive a goal and decide which steps to take to achieve it: they query data in the ERP or CRM, prioritise and propose actions, with a person approving anything important. ChatGPT and other general assistants are very good at conversation and drafting, but they do not know your systems or run a process from start to finish unless you connect them.
How much does AI consulting for a business cost?
There is no standard price for AI consulting for a business: it depends on the number of processes, the systems that need connecting, the quality of your data and the level of oversight you need. Be wary of fixed quotes given before anyone has seen the process. The sensible start is mapping a single process, which tells you what to automate and what not to before you invest more.
Is AI training mandatory for businesses?
There is no specific mandatory course, but since 2 February 2025 Article 4 of the EU AI Act has required organisations that use AI systems to take AI literacy measures for their staff, taking into account their knowledge and how the systems are used. The European Commission clarifies that no certificate is needed: role-based training and an internal record are enough.
Is there public support for AI for small business in the EU and the UK?
In the EU, European Digital Innovation Hubs act as one-stop shops offering test-before-invest services, training and financing advice, with a reinforced AI focus since 2025. In the UK, the government announced on 28 January 2026 free AI foundations courses for every adult through its AI Skills Hub, aiming to reach at least 2 million SME employees. In Spain, Kit Digital and Kit Consulting had no new open calls as of 12 October 2026.
Where should a small business start with AI?
With a specific, measurable process, not with a list of the best AI tools for business. Choose a repetitive task that costs hours today, measure how it is done, map it step by step and use AI only where it adds judgement, with a person approving. With the MAPA method (map, analyse, pilot and audit), nothing is automated without an approved map and nothing goes live without a pilot.
References
Sources
Checked and verified on 12 October 2026.
- Eurostat, Use of artificial intelligence in enterprises (data extracted in December 2025).
- Eurostat, Artificial intelligence by size class of enterprise (isoc_eb_ai) (updated 15 June 2026).
- INE, Survey on the Use of ICT and e-commerce in enterprises. Year 2024 – First quarter 2025. Definitive data (22 October 2025).
- INE, ICT in enterprises (first quarter of 2025): artificial intelligence, by activity and size, in Spanish (22 October 2025).
- INE, Survey on the use of ICT and e-commerce in enterprises: latest data, in Spanish (checked on 12 October 2026).
- Banco de España, Adoption of artificial intelligence in Spanish firms: an initial analysis based on the Banco de España Business Activity Survey, Economic Bulletin 2025/Q2 (7 May 2025).
- EUR-Lex, Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (AI Act), Articles 2, 4 and 50 (OJ, 12 July 2024).
- EUR-Lex, Regulation (EU) 2026/1744 (Digital Omnibus on AI) (8 July 2026).
- European Commission, AI Literacy: Questions & Answers (updated 27 July 2026).
- European Commission, European Digital Innovation Hubs (updated 19 December 2025) and the EDIH catalogue.
- GOV.UK, Free AI training for all as government and industry programme expands to provide 10 million workers with key AI skills by 2030 (28 January 2026).
- BOE, Order TDF/38/2026 (Kit Consulting) and Order TDF/39/2026 (Kit Digital), in Spanish (28 January 2026).
- Fundae, FAQs: company training credit, in Spanish (checked on 12 October 2026).
This article was created with the help of AI and reviewed by José Galán. I take great care over every post and every translation, but the odd mistake can still slip through. If you find one, write to me: you will be helping me improve.
Custom AI agents
Your first process with AI? Let’s map it together.
I map one of your company’s processes with the MAPA method and design a custom AI agent that analyses your data, prioritises and suggests what to do. Data stays in the EU and you approve every important action. If a simple automation is enough, I will tell you that too.
See how I design AI agents