1. Where a chatbot makes sense in B2B
B2B websites have different visitors from online shops. Anyone looking for marketing consultancy, a machine or a software solution rarely wants to buy straight away, but first wants to understand whether the provider is a good fit. This gives rise to a number of use cases in which a chatbot can add real value:
- Pre-qualification: the bot asks about industry, company size and requirements and suggests a suitable appointment.
- Recurring factual questions: delivery times, areas of application, technical requirements, how a collaboration works.
- Orientation on large websites: the bot leads to the right pages, documents or contact persons.
- Availability outside business hours: requests are recorded and dealt with on the next working day.
A chatbot makes little sense, on the other hand, on small websites with few services, for very individual consulting topics, or when nobody has time to maintain it. A bot that answers with outdated information does more harm than a clearly visible contact form.
Depending on how it is used, the same technology acts as a service or as an obstacle.
2. What Article 50 of the AI Act requires
Article 50(1) of the AI Act (Regulation (EU) 2024/1689) states, in essence: providers shall ensure that AI systems intended to interact directly with natural persons are designed in such a way that the persons concerned are informed that they are interacting with an AI system – unless this is obvious from the point of view of a reasonably well-informed, observant and circumspect person, taking into account the circumstances.
Paragraph 5 adds: the information must be provided in a clear and distinguishable manner at the latest at the time of the first interaction and must conform to the applicable accessibility requirements.
Who is obliged?
The obligation under paragraph 1 is addressed to the provider. According to Article 3 of the Regulation, a provider is anyone who develops an AI system or has one developed and places it on the market or puts it into service under its own name or trademark. Anyone who merely uses a ready-made chatbot product is usually a deployer. The line can be blurred, however, for example when a company assembles a bot itself on the basis of a language model and uses it under its own name.
In practice, the distinction is less important than it sounds: a short, clear notice at the start of the conversation – such as “You are chatting with an AI assistant” – generally fulfils the requirement and is good practice anyway. If you give the bot a human-sounding name and photo, take particular care that the notice does not get lost.
Since when has the obligation applied?
Under Article 113, the transparency obligations under Article 50 have applied since 2 August 2026. The amending Regulation (EU) 2026/1744, the so-called Digital Omnibus on AI, which entered into force in July 2026, mainly postponed the deadlines for high-risk systems and introduced a transitional period for the machine-readable marking of artificially generated content (paragraph 2) for systems already offered before 2 August 2026. To the best of our knowledge, the notice for direct interaction under paragraph 1 was not postponed. As of: October 2026.
This section is an assessment, not legal advice. For your specific case, please consult a lawyer.
Irrespective of the AI Act, the GDPR applies: chat logs often contain personal data. Check where the provider processes the data, whether a data processing agreement is in place and how long conversations are stored.
3. The handover to humans
In B2B, the actual enquiry almost always arises in a conversation with a human. A chatbot is therefore above all only as good as its handover. Three points are crucial:
- Clear trigger: on certain signals – a question about price, a quote or an appointment, dissatisfaction, repeated questions – the bot actively offers contact with a human.
- No loss of information: the conversation so far is handed over too, so the prospect does not have to repeat anything.
- Honest expectations: if nobody is available at the moment, the bot says so and states when a reply will come.
This is how a well-configured chatbot leads a prospect to a personal conversation.
4. How to measure the benefit
A chatbot should be measured by the same figures as any other marketing tool. Useful ones are:
- Share of visitors who start a conversation
- Share of conversations that end with a handover to a human
- Number of qualified enquiries from the chat
- Comparison: have enquiries via form and phone developed differently since the bot went live?
Set up dedicated events for conversation start and handover in your web analytics and record in the CRM that an enquiry came from the chat. Also read conversation logs regularly: they show which questions your website has not yet answered – a valuable source of new content.
5. Checklist
- Define a specific use case before choosing a tool.
- Display the AI notice clearly and accessibly at the start of every conversation.
- Do not automatically overlay the bot on the content; offer it unobtrusively.
- Limit answers to your own well-maintained content.
- Set up the handover to humans with clear triggers.
- Check data processing and the retention period of chat logs.
- Create events for conversation start and handover in your web analytics.
- Read conversation logs monthly and improve content.
6. Conclusion
In B2B, a chatbot can pre-qualify enquiries and improve availability, provided it is honestly labelled, well maintained and hands over to a human in good time. However, it replaces neither an understandable website nor personal sales. For small websites with a manageable range of services, a clearly visible contact form with a quick response is often the better and cheaper solution.
7. Frequently asked questions
Do I have to label my website chatbot as AI?
Under Article 50(1) of the AI Act, users must be informed that they are interacting with an AI system, unless this is obvious. The obligation is addressed to the provider of the system; a clear notice at the start of the conversation is advisable in any case. This is not legal advice.
Since when has the labelling obligation for chatbots applied?
The transparency obligations under Article 50 have applied since 2 August 2026. To the best of our knowledge, the Digital Omnibus on AI has not postponed the notice requirement for direct interaction.
Does a chatbot bring more enquiries in B2B?
It depends on how it is used. On websites with many recurring questions and a clear handover to sales, it can help; on small sites it is often more of a nuisance. Measure conversation starts, handovers and qualified enquiries.
What does a website chatbot cost?
The range extends from features included in existing website or CRM tools to custom-developed solutions. In addition to the licence, you should factor in the ongoing maintenance of the content, because it determines the quality of the answers.
Can a chatbot have a human name?
A name is not prohibited, but it increases the risk that users take the bot for a human. That makes a clear notice at the start, saying that an AI system is responding, all the more important.
