In short: A quarter of businesses in Germany with at least ten employees used AI in 2025, one in five among small and medium-sized firms, and in a Bitkom survey of companies with 20 or more employees a majority said for the first time in 2026 that they use AI. AI is most often used in customer contact. On the phone, however, it is only slowly arriving: only about one business in ten uses speech recognition. This post shows where businesses stand on digital transformation in 2026, what holds them back and what voice AI actually changes on the phone, with a source for every figure.
We rewrote this post in October 2026. The first version from February 2026 promised response times in the millisecond range thanks to edge computing, announced the cloning of employees' voices and real-time emotion detection, and ended with a quote without a source. These statements could not be backed up or did not describe what EchoCall offers. We have removed them. We now also test the claim from the old title, that voice AI is "the new internet", against the figures.
This post is an overview of where things stand. Which developments shape customer service in 2026 is described in Customer Service Trends 2026: Eight Shifts in Conversational AI, Voice AI and AI Agents, with Sources. How to introduce AI agents is covered in AI Strategy 2026: A Practical Guide to AI Agents in Customer Service, from Contact Analysis to Rollout. What an AI phone agent costs and how to choose one is explained in AI Phone Agent 2026: What It Costs, How It Works and How to Roll It Out.
Digital transformation 2026 in figures: how many businesses use AI
Official statistics are the most cautious. According to the Federal Statistical Office of Germany (Destatis), just over a quarter of businesses in Germany with at least ten employees used AI in 2025 (26%). Among small businesses with 10 to 49 employees the figure was 23%, among medium-sized businesses with 50 to 249 employees 36% and among large businesses with 250 or more employees 57%. [1]
This puts Germany ahead in a European comparison. Eurostat reports 25.97% for Germany in 2025, up from 19.75% in 2024. The EU average rose from 13.48% to 19.95%. [2] Denmark (42.03%), Finland (37.82%) and Sweden (35.04%) are at the top, Romania (5.21%), Poland (8.36%) and Bulgaria (8.55%) at the bottom. [3]
A survey by the digital association Bitkom comes out much higher. In summer 2026, 57% of 603 surveyed companies with 20 or more employees said they use AI, up from 36% a year earlier and 20% two years earlier. Another 38% are planning or discussing it. At the same time most see themselves only at the beginning. Of the companies using AI, 28% say they make little use of its potential and 59% say they make no use of it at all. [4]
| Source | Who was counted | Period | AI use |
|---|---|---|---|
| Destatis | Businesses with 10+ employees | 2025 | 26% |
| Eurostat | Businesses with 10+ employees, EU | 2025 | 19.95% |
| KfW SME Panel | SMEs, including micro businesses | 2022 to 2024 | 20% |
| Bitkom | Companies with 20+ employees | Summer 2026 | 57% |
The figures do not contradict each other, they count differently. Bitkom asks later and only companies with 20 or more employees. Official statistics cover businesses with ten or more employees in an EU-wide harmonised survey, which is why the Destatis and Eurostat figures for Germany match. KfW also counts micro businesses and asks about a three-year period. For your planning this means that AI is widespread among larger firms in 2026, but not yet among small businesses.
AI in small and medium-sized businesses: micro firms and sectors
The official figures above leave out micro businesses with fewer than ten employees. KfW, the German development bank, fills this gap with its SME Panel. According to the panel, 20% of small and medium-sized businesses used AI between 2022 and 2024, almost 780,000 businesses. Between 2016 and 2018 it was 4%. Among businesses with fewer than five employees the share was 19%, with 5 to 9 employees 20%, with 10 to 49 employees 29% and with 50 or more employees 36%. [5]
The sector also makes a difference. Among German SMEs, knowledge-intensive services use AI most often (28%), followed by research-intensive manufacturing (23%). In construction it is 8%. [5] Eurostat shows the same pattern across the EU: information and communication leads with 62.52%, followed by professional, scientific and technical activities with 40.43%, while construction is at 10.79%. [3]
What matters more than size is the attitude to digitalisation. According to KfW, SMEs with a digitalisation strategy use AI with a probability of 35%, businesses without any digitalisation projects with 19%. [5] This suggests starting with AI where processes already run digitally.
What holds businesses back on AI
Destatis asked businesses that do not yet use AI but had considered it about their reasons. That applied to one in five businesses without AI. The most frequent reasons were lack of knowledge (72%), uncertainty about the legal consequences (62%) and data protection concerns (60%). These were followed by incompatibility with existing systems (45%), problems with the availability or quality of data (44%), cost (32%) and ethical considerations (23%). [6]
So cost is not the biggest obstacle, knowledge, law and data protection are. All three can be clarified before an introduction. What the GDPR requires for voice agents and chat is described in AI and Data Security 2026: The GDPR Guide for Voice Agents and AI Chat in Customer Service. Which requests are suitable to start with and how to bring your team along is covered in the strategy guide.
What businesses use AI for: customer contact first
According to Bitkom, AI is used most often in customer contact in 2026, for example to handle enquiries (72% of companies using AI), followed by marketing and communication (54%). Internal knowledge management (22%, up from 11% the year before) and AI in sales (14%, up from 5%) have grown quickly. [4]
AI agents are coming into focus next, meaning systems that act on their own instead of only answering. Of the companies using or planning AI, 11% already use such agents, 29% are planning to and 31% are discussing it. Bitkom president Ralf Wintergerst names three prerequisites: reliable data, connected systems and clear rules on what an agent may do. [4]
One difference stands out in the type of AI. Among German businesses that use AI, 52% use text mining, 42% speech recognition and 35% natural language generation. [1] Measured against all businesses, that is 10.9% for speech recognition and 9.02% for speech generation. [2] Spoken language has therefore reached about one business in ten.
The customer side: AI has arrived in everyday life
Customers are moving faster. According to the ARD/ZDF media study 2026, 59% of people in Germany aged 14 and over use AI tools at least occasionally, 92% of 14 to 29 year olds and 15% of those aged 70 and over. More than half (54%) treat the results with scepticism, though. 81% consider source references important or very important, but only 43% of users find them transparent. [7]
For customer service this means that many customers know AI and expect fast answers. But they want to understand where an answer comes from and to reach a human when needed. Language is another factor. Around 77% of people in Germany spoke only German at home in 2024. 15.5 million mainly or exclusively used another language at home, most often Turkish, Russian and Arabic. [8] A phone agent that recognises the caller's language reaches these customers more easily than a German-only phone menu.
The phone: important, but behind on AI
The phone remains a central channel. In a Gartner survey of 199 customer service leaders, they named live chat and the phone as the most valuable channels today. In two years they expect GenAI chatbots in first place, followed by live chat and GenAI voicebots. [9] So the phone is not being replaced, it is getting an AI layer, more slowly than chat.
At the same time many people do not like phoning. According to Bitkom, 36% of Germans have put off necessary calls out of fear of phoning, and 44% of 16 to 29 year olds. 33% feel uncomfortable phoning strangers, and 32% prefer to make appointments by email or via a booking platform. 40% would rather write a message than call, and 52% of younger people. [10]
For businesses this creates a double task. People who call expect a fast answer. People who would rather not call need an equivalent route in chat. Both routes should draw on the same content.
What voice AI changes on the phone: from phone menu to conversation
Many hotlines still use a phone menu (IVR): "Press 1 for appointments." An AI phone agent, by contrast, understands freely spoken sentences, answers from a stored knowledge base and can carry out actions during the call, such as entering an appointment in the calendar.
| Feature | Phone menu (IVR) | AI phone agent (voice AI) |
|---|---|---|
| Input | Keys or single words | Freely spoken sentences |
| Answers | Fixed announcements | Formulated answers from your knowledge base |
| Actions | Transfer to a queue | e.g. book an appointment, record details, create a ticket |
| Languages | Usually one or two | Several, depending on the provider with automatic language detection |
| When it does not fit | Back to the menu | Transfer to a person |
| Maintenance | Record new announcements | Update the knowledge base |
An AI phone agent does not run itself. It answers only as well as the content you provide, and it can make mistakes. Since 2 August 2026, under Article 50 of the AI Act, callers must also be informed that they are talking to an AI, unless this is obvious. [11] With EchoCall you write this notice into the phone agent's first message yourself.
The EchoCall voice agent understands more than 30 languages and can detect the caller's language automatically. You fill the knowledge base with documents (PDF, TXT, Markdown, CSV, JSON), individual web pages, a crawl of your whole website or free text. If a question does not fit, the agent transfers the call to a phone number. What a phone agent costs in operation is worked out in the guide to AI phone agents.
Is voice AI the new internet?
No, and we no longer claim it is. The figures paint a more sober picture. In 2025 a quarter of businesses in Germany used any AI at all, and about one in ten used speech recognition. [1] [2] Customer service leaders expect voice AI to be among their three most valuable channels in two years, but behind chatbots and live chat. [9] Voice AI therefore changes one particular channel, the phone, and does so where many similar calls come in: appointments, opening hours, order status, callback requests.
How quickly this spreads depends less on the technology than on the obstacles Destatis measures: knowledge, law and data protection. [6] Businesses that clarify these three points can start small today, for example with an agent for calls outside opening hours.
You can hear how an agent sounds with the content of your website in our live demo: enter your website, and after about 20 seconds call or write to it in the browser, without signing up. After that there is a trial: 30 voice minutes and 30 chat conversations for 14 days, and the credit card is verified with €1. The voice agent costs €0.30 per minute with no setup fee, and phone numbers and telephony come on top. All plans are on the pricing page.
Frequently asked questions about digital transformation 2026
How many businesses in Germany use AI?
According to Destatis, 26% of businesses with ten or more employees used AI in 2025, and 57% of large businesses. [1] In a Bitkom survey of companies with 20 or more employees, 57% said in summer 2026 that they use AI. [4] Among SMEs including micro businesses, the figure was 20% according to KfW. [5]
What holds back digital transformation in SMEs?
Businesses that considered AI but did not introduce it mainly cite lack of knowledge (72%), legal uncertainty (62%) and data protection concerns (60%). Cost only follows with 32%. [6]
What is the difference between voice AI and a phone menu?
A phone menu (IVR) works with fixed announcements and keys. Voice AI understands freely spoken sentences, answers from a knowledge base, can carry out actions such as booking an appointment and transfers to a person when it gets stuck.
How fast does an AI phone agent respond?
We deliberately do not give a fixed number of milliseconds. The response time depends on the phone connection, the length of the answer and whether the agent queries another system during the call, such as a calendar. The best way is to check it yourself: in the live demo you can call an agent with the content of your website in the browser.
Which languages can an AI phone agent speak?
That depends on the provider. The EchoCall voice agent understands more than 30 languages and can detect the caller's language automatically. This matters in Germany, where 15.5 million people mainly or exclusively use a language other than German at home. [8]
Is voice AI the new internet?
No. Voice AI mainly changes the phone as a channel. Customer service leaders expect voice AI to be among their three most valuable channels in two years, but behind chatbots and live chat. [9]
Sources
Sources marked "(in German)" are in German. We checked the Gartner figures via a verbatim reprint, because Gartner blocks automated access to its website.
- Federal Statistical Office of Germany (Destatis), businesses using artificial intelligence technologies by size class, as of 24 November 2025 (in German): destatis.de
- Eurostat, database isoc_eb_ai, Artificial intelligence by size class of enterprise: ec.europa.eu
- Eurostat, Statistics Explained, Use of artificial intelligence in enterprises, 2025 data: ec.europa.eu
- Bitkom, press release of 14 September 2026, survey of 603 companies with 20 or more employees (in German): bitkom.org
- KfW Research, Fokus Volkswirtschaft No. 533 of 11 February 2026, AI use in SMEs (in German): kfw.de
- Federal Statistical Office of Germany (Destatis), survey on ICT use in enterprises, 2025 results (in German): destatis.de
- ZDF press portal, ARD/ZDF media study 2026 of 22 September 2026 (in German): presseportal.zdf.de
- Federal Statistical Office of Germany (Destatis), figure of the week of 17 February 2026, microcensus 2024 results on languages spoken at home (in German): destatis.de
- Gartner, survey of 199 customer service leaders (April and May 2026), reprinted by IT-Online on 28 August 2026: it-online.co.za
- Bitkom, press release of 11 April 2025, survey of 1,004 people aged 16 and over (in German): bitkom.org
- Regulation (EU) 2024/1689 (AI Act), Articles 50 and 113: eur-lex.europa.eu
All links were checked on 6 October 2026. This post is not legal advice.
Read next
- Customer Service Trends 2026: Eight Shifts in Conversational AI, Voice AI and AI Agents, with Sources
- AI Strategy 2026: A Practical Guide to AI Agents in Customer Service, from Contact Analysis to Rollout
- The Rise of AI Voice Agents 2025 to 2026: What Really Changed in Technology, Prices, Funding, Law and Analyst Forecasts
- AI Voice Agent Statistics 2026: Market Size, Adoption, Cost per Minute, Highest-Value Niches and the Law, Every Number Sourced
- AI and Data Security 2026: The GDPR Guide for Voice Agents and AI Chat in Customer Service
