AI chatbot for your website.

What is an AI chatbot? Definition and how it works

An AI chatbot is software that conducts text-based conversations with website visitors or app users, based on a Large Language Model (LLM) instead of fixed rules. Unlike bot frameworks of the 2010s, no intent tree needs to be configured, the LLM understands requests directly. How it works: user input → LLM inference with context (knowledge base + conversation history) → structured response + optional tool call (book appointment, create CRM entry, escalate). The typical DACH setup configuration: Microsoft Azure OpenAI Service as primary LLM (gpt-4o, EU Data Boundary), Qdrant as vector DB for knowledge base embeddings, Hetzner Frankfurt as hosting region. Response latency in chat window typically 800-1,500ms, comparable to human response times.

AI chatbot vs. classic live chat, when does what fit?

Classic live chat tools need human staff. Advantage: emotional empathy, situational judgment, complex escalations. Disadvantage: business hours only, high personnel cost factor, sluggish scaling at traffic spikes. AI chatbots are superior in four aspects: 24/7 without surcharge, learn from your knowledge base, answer hundreds of requests in parallel without wait time, integrate natively with appointment booking and CRM. Hybrid models (AI chatbot for standard requests, human for complex escalations) are common in DACH SMBs. Concrete automation rate depends on use case and knowledge base. Rule of thumb: at more than 100 website visitors per day with chat engagement rate above 5%, the AI chatbot pays off from month one.

How do you choose an AI chatbot provider in DACH 2026?

The DACH AI chatbot market is fragmented. Three segments: first, international live chat platforms with AI add-ons. Second, pure AI chatbot specialists. Third, open-source frameworks for self-hosting. Differentiation across five dimensions: EU sovereignty (hosting, LLM region), GDPR compliance (DPA, AI Act disclosure), industry depth (templates), integration breadth (CRM, appointment, helpdesk), pricing model. Check concretely before signing: Where do chat histories and LLM inference run? Is there a DPA per Art. 28 GDPR? Is AI Act disclosure a default or a configuration effort? Rinqo positions itself as a full-EU-sovereign multi-agent platform: phone, chat and email agent plus a personal assistant, all from one login and sharing the same knowledge base.

AI chatbot GDPR-compliant, what do I need to know?

Three GDPR requirements are mandatory when using AI chatbots: first, server location EU for chat history, knowledge base, LLM inference. Rinqo uses Hetzner Falkenstein and Nuremberg. Second, cookie-free or consent-compliant tracking, Rinqo chat widget is cookie-free by default, tracking only with explicit consent. Third, DPA contract with sub-processor handover clause (LLM provider). Rinqo provides the DPA automatically before contract conclusion. AI Act Art. 50 adds: disclosure obligation from 02.08.2026, at chat start it must be clearly recognizable that the user is talking to an AI (chat header disclaimer 'You are chatting with the AI assistant of [company]').

AI chatbot costs 2026, what do you really pay?

Three pricing models exist in the market: per-conversation (every conversation billed individually, often with a separate AI add-on), per-seat (costs scale linearly with your team) and flat-rate. Rinqo bills flat plus credits: Starter €99 / Pro €349 / Business €799 net per month, Enterprise from €1,499. One credit equals 1 phone minute, 5 messages or 2 emails; Starter includes 500 credits, Pro 2,000, Business 6,000. Every plan includes phone, chat and email agent plus the personal assistant from one platform, no per-channel add-on surcharge. Setup effort at Rinqo: none, the widget is embedded in 10 minutes. Before signing anywhere, calculate your actual conversation volume — per-conversation models often tip over at a few hundred chats per month.

What is a chatbot? The short answer

A chatbot is software that holds text-based conversations. First generation (1990s–2010s): rule-based, script-driven, broke whenever the wording left the script. Second generation (2017–2022): intent trees with classic NLU, better language understanding but maintenance-heavy. Third generation (since 2023): LLM-based, understands any wording, answers from a knowledge base, no script. Rinqo's AI chatbot is third generation: a Large Language Model (Azure OpenAI inside the EU Data Boundary) plus your knowledge base (website scraping, FAQ, documents). When you search for 'what is a chatbot' online, run the 30-second test: ask the typical question of your customer base in two different wordings — if both get answered correctly, it is a modern AI chatbot. If not, it is a legacy rule-based system.

• Generation 1 (rule-based): breaks at any wording outside the script
• Generation 2 (intent tree): better language understanding, high maintenance
• Generation 3 (LLM + knowledge base): understands any wording, learns from your content
• Rinqo's AI chatbot is Generation 3, EU-sovereign hosted

Free chatbot vs. Rinqo, when does each make sense?

'Free chatbot' is a valid first question. The answer: free tiers exist at several providers, typically capped at 50–100 conversations per month, plus open-source frameworks for self-hosted setups — and they work as long as your volume stays small. What you don't get on a free tier: a dedicated DPA per Art. 28 GDPR, an EU-sovereign LLM stack without US cloud exposure, AI Act Art. 50 default disclosure, a multi-channel knowledge base. For private projects or a hobby shop, that's fine. For SMB customer service with GDPR obligations (healthcare, legal, education, B2B), things get expensive fast once request volume goes above 50–100/month — both in tier jumps and in compliance risk. Rinqo starts at €20 prepaid (no subscription required), is GDPR-default and covers phone, chat and email, plus a personal assistant for your team. If you want to test volume without a free tier: prepaid model instead of free trial.