Service · AI Chatbot
AI Chatbot for Sales and Customer Service
An ArkeonTech AI chatbot answers customer enquiries on your website, Instagram and Facebook Messenger within seconds, qualifies leads automatically and hands hot prospects to your team. At a fixed price after a free initial call. Live in 2-4 weeks, GDPR-compliant on EU servers.
Last updated:
- fixed price Model
- tailored Contract
- 2-4 weeks Live in
What does an AI chatbot cost for an SMB?
Costs depend on scope, channels and integrations. At ArkeonTech you receive a transparent fixed-price offer after a free initial call: a one-off setup for concept, build and integration plus a monthly service for operation, hosting, model updates and fine-tuning. Contract term and notice are agreed individually in the offer.
| Service | Included |
|---|---|
| Setup (one-off) | Use-case concept, knowledge base, integration, testing, go-live |
| Operation (monthly) | EU hosting, monitoring, model updates, answer fine-tuning |
| Initial call & demo (free) | 20-minute potential analysis with a concrete recommendation |
What is the difference between a chatbot and an AI agent?
A classic chatbot follows rigid click flows and fails on free-form questions. An AI agent understands natural language, keeps conversational context, accesses your product and company data and takes action - booking appointments, filling lead forms or escalating to a human.
- Context awareness across the whole conversation instead of keyword matching
- Access to your knowledge base: products, pricing, opening hours, processes
- Actions, not just answers: bookings, CRM entries, sales handover
- Clean escalation to humans when a case gets complex
Which channels does the AI chatbot cover?
One agent, all channels: website chat, Instagram DMs, Facebook Messenger, WhatsApp Business API, SMS and email. CRM integration (HubSpot, Salesforce, Pipedrive, Notion, Airtable) is part of the standard setup; other systems connect via REST APIs or webhooks.
Which companies is an AI chatbot suitable for?
Companies with 5 to 250 employees that regularly receive similar enquiries via website or social media. Typical examples are online retail, medical practice groups, estate agencies and financial advisory firms. What decides it is not headcount but enquiry volume: where questions only arrive occasionally, maintaining the knowledge base costs more time than the agent saves. We tell you how we read your case in the free initial call. At the upper end our fit ends where a corporation has its own IT department with its own operating model and procurement process.
- Company size: 5 to 250 employees, focus on 10 to 100
- Regularly recurring enquiries via chat, form or DM
- Typical sectors: e-commerce, healthcare, real estate, financial advice, hospitality, industry
- No in-house development team required - we run the operation
When is an AI chatbot the wrong solution?
In three cases we advise against it. First, when the answers are documented nowhere and nobody wants to write them down: an agent reproduces what the knowledge base contains, and without that foundation the project fails on editorial work, not on technology. Second, for answers that are individual professional judgements - legal advice, medical diagnosis, binding commitments where every quote is calculated individually. The agent records such requests and passes them on; it must not decide them. Third, when your customers almost exclusively call: then an AI phone assistant is the right channel and a chat window simply stays unused.
Can an AI chatbot be GDPR-compliant?
Yes. ArkeonTech hosts exclusively on European servers, signs data processing agreements with all sub-providers and documents every data flow. Personal data is stored only as long as its purpose requires - the setup passes any privacy review.
Which regulations apply to an AI chatbot?
Four sets of rules almost always apply. Since 2 August 2026 the EU AI Act has required the chatbot to identify itself as AI. The GDPR requires users to be informed and a data processing agreement with every provider that processes chat data on your behalf. The German TDDDG governs what the chat window may store in the browser, and where there is a works council, it has a say as soon as the data allows employees' work to be evaluated. What the chatbot says counts as the company's own statement according to a judgment of the Higher Regional Court of Hamm of 12 May 2026, and if it rejects applications itself, Article 22 GDPR only permits this in exceptional cases and requires a route to a human. From 9 December 2026, Germany's new Product Liability Act applies as well, under which software counts as a product. The overview shows what usually applies; it does not replace legal advice for your specific case.
| Regulation | What it requires | What it means for your chatbot |
|---|---|---|
| Art. 50 para. 1 AI Act | People must be able to tell that they are communicating with an AI; applies since 2 August 2026 and binds the provider, which includes a company that has the chatbot developed and uses it under its own name | A notice in the first chat message; according to the European Commission's guidelines, a sentence in the legal notice, terms or privacy policy, calling it an assistant or showing a human photo is not enough on its own |
| Art. 13 GDPR | Information on the purpose, legal basis and recipients of the data when it is collected | A privacy notice in the chat window and a section in the privacy policy |
| Art. 28 GDPR | A data processing agreement with every provider that processes chat data on your behalf | Agreements for hosting and the language model; if anything runs outside the EU, the transfer needs its own legal basis (Art. 44 et seq. GDPR) |
| Art. 35 GDPR | A data protection impact assessment before launch where processing is likely to result in a high risk; the list of Germany's Data Protection Conference names customer support using AI as a typical field (no. 11) | A duty of the company deploying the chatbot; the service provider must support it (Art. 28(3)(f) GDPR), for example with a description of data flows and safeguards |
| § 25 TDDDG | Storing or reading information in the browser only with consent, unless strictly necessary for the requested service | If the chat window stores more than the chat technically needs, for example for analytics, it needs consent |
| § 87 para. 1 no. 6 BetrVG | Works council co-determination for technical systems capable of monitoring employees' conduct or performance | Applies, for example, when handovers to staff can be evaluated; involve the works council before launch |
| Art. 22 GDPR, from 20 November 2026 also § 37a BDSG | No decision based solely on automated processing with legal or similarly significant effect, unless it is necessary for a contract, permitted by law or based on explicit consent; in the contract and consent cases the customer can obtain human intervention, express their point of view and contest the decision; if it relies on a score about future behaviour, such as payment default, § 37a BDSG gives the right, from 20 November 2026, to contest it, to express one's point of view and to a decision by a human | Information such as delivery status or opening hours is not such a decision; if the chatbot itself rejects applications or terminates contracts, this is only permitted under these exceptions and requires a route to a human who reviews the decision |
| § 5 UWG | No misleading statements, for example about qualifications; according to the Higher Regional Court of Hamm, the answers of an AI chatbot are commercial practices of its operator, even if the bot had only been trained on correct data (judgment of 12 May 2026, 4 UKl 3/25, final) | What the chatbot says counts as a statement by your company; check the questions customers will foreseeably ask for false answers before launch and, where a correct answer cannot be ensured, store fixed answers or hand over to a human |
| Art. 246b § 3 para. 3 EGBGB | If a company uses online tools such as chatbots for financial services at a distance, the consumer can demand human intervention before the contract is concluded and, in justified cases, afterwards; applies since 19 June 2026 | Affects financial service providers, for example for accounts, loans, investments or payments: a human must be reachable behind the chatbot on request |
| Product Liability Act from 9 December 2026 | Software counts as a product; if a defect in the product causes personal injury or damage suffered by natural persons to property not used exclusively for professional purposes or to data not used professionally, the manufacturer is strictly liable, and this cannot be excluded by contract towards injured parties (sections 1 and 18) | The manufacturer is whoever develops the chatbot or has it developed, and, unlike under the AI Act's disclosure duty, also whoever merely presents themselves as the manufacturer with their own name or brand (section 3); damage within your own business remains a matter of contract; chatbots placed on the market or put into service by 8 December 2026 inclusive remain under the previous act (section 23) |
In depth in our articles:
Do you build the chatbot on site?
Yes, where it helps. ArkeonTech is based in Aalen in Baden-Württemberg, southern Germany. The demanding part of a chatbot project is not the technology but the knowledge base: which questions actually arrive, which answer applies, who decides the exceptions. Two hours around a table settle that faster than five rounds of email. Companies within about an hour of Aalen get that workshop in person; everything after it runs remotely, as do projects anywhere else in Germany and beyond.
Who is your point of contact?
Michael Kaiser, owner of ArkeonTech and the engineer who builds your chatbot. Business informatics (DHBW Heidenheim), B.Sc. Business Analytics (Aalen University), previously several years as an ERP consultant to mid-sized companies in Baden-Württemberg. The initial call, the knowledge-base build and the later fine-tuning all run through the same person - there is no sales or account layer in between.
Email: m.kaiser@arkeontech.de
FAQ
Frequently asked questions
Standard projects go live in 2-4 weeks: week 1 concept and knowledge base, weeks 2-3 integration and testing, then go-live with ongoing fine-tuning.
No. ArkeonTech handles hosting, monitoring, model updates and answer fine-tuning in the monthly service plan. Your team only works with the results, e.g. qualified leads in the CRM.
The agent recognises its limits and escalates cleanly to a human - via email, Teams/Slack message or ticket, with the full conversation handed over.
Yes. The agent books directly into your calendar (Calendly, Google Calendar, Outlook) and writes the lead data into your CRM in parallel.
Yes. We review real conversations, extend the knowledge base and adjust answers monthly, so the resolution rate keeps rising without any effort on your side.
Yes, as part of the monthly service: monitoring, model updates and answer tuning based on real conversations. Clients near our Aalen office can add a quarterly on-site review if they want one.
Find out in 20 minutes whether an AI chatbot pays off for you
Free initial call: we analyse your use case and give you a realistic effort/benefit estimate.
- No obligation
- 20 min call
- GDPR compliant