AI agents for your business.

What is an AI agent?

An AI agent is a software system that completes tasks on its own instead of merely answering questions. It understands a request, decides on the next steps and carries them out. Four traits set it apart from a classic chatbot. First, tool use: the agent independently accesses tools such as a calendar, a CRM or a knowledge base. Second, multi-step capability: it breaks a request down into several steps and works through them one after another. Third, memory: context is retained throughout the conversation, so nobody has to repeat themselves. Fourth, escalation logic: the agent recognises its limits and hands defined cases over to humans. An example makes it concrete. A caller says: “I need an appointment next week, ideally in the morning.” A chatbot reads out the opening hours. An AI agent checks the calendar, offers free slots, books the chosen one and sends your team an email summary. Technically, an AI agent consists of three building blocks: a language model as the understanding and decision layer, declared tools (function calling) and a knowledge base holding your company information. At rinqo, the language model runs contractually guaranteed within the EU, with no training on your data. The knowledge base is built through website import and PDF upload, and answers include source citations. That way you can always trace what the agent based its answer on.

AI agent examples from everyday SME life

Five use cases cover most of everyday SME life: phone reception, chat leads, email triage, an internal assistant and proactive follow-up. First, phone reception: the agent answers every call, clarifies the request, books appointments straight into the calendar and escalates emergencies to your team immediately. After every call, an email summary follows. Second, chat on your website: the widget answers questions from your company knowledge and qualifies prospects before your sales team takes over. Third, email triage: the email agent works with your existing inbox via IMAP/SMTP, classifies incoming enquiries and answers routine cases. Fourth, the personal agent: your team queries company knowledge in a chat, from processes and prices to product details, fed from the shared knowledge tree. Fifth, proactive follow-up: appointment reminders and call-backs on open enquiries before the lead goes cold. Your team takes over the cases that really need people.

• Phone reception: answer the call, book the appointment, escalate emergencies, email summary
• Chat leads: answer website questions, qualify prospects
• Email triage: classify the inbox, send routine replies (IMAP/SMTP)
• Internal assistant: company knowledge in a chat, one knowledge tree for all agents
• Proactive follow-up: appointment reminders and call-backs before leads go cold

Building an AI agent: the platform route vs the framework route

There are two routes to your own AI agent: a platform we set up for you or building it yourself on an open-source framework. The platform route with rinqo runs by us: create an account, have your website imported, upload PDFs if needed, assign a phone number or embed the chat widget, run a test. After that, the agent knows your business and answers with source citations. You can connect calendars (Google, Outlook or CalDAV) yourself, plus your own systems via REST API and webhooks. CRM and accounting (HubSpot, sevdesk, lexoffice) or project tools (Linear, Jira) we connect on request. The framework route is the opposite approach: maximum flexibility, but with the full operational burden. You need hosting, deployment, monitoring, telephony integration, speech synthesis, a vector database and someone to maintain it all permanently. On top of that comes the compliance work: DPA, EU hosting, AI Act labelling, logging. That is a developer project with an ongoing DevOps commitment, not a weekend setup. For software teams with special requirements it can be worth it. For a small or medium-sized business that wants to take the load off phone, chat and email, the platform is the faster route. The sober maths: our setup versus weeks of building.

Types of AI agents: by channel and role

AI agents differ along two dimensions: channel and role. By channel there are three basic types. Voice agents hold phone conversations in real time, the most technically demanding form, because every delay is audible. At rinqo, telephony runs through EU data centres, and the speech synthesis comes from Germany and is EU-hosted. Chat agents sit on your website as a widget and reply in writing. Email agents work asynchronously in your inbox via IMAP/SMTP. By role, there are customer-facing and internal agents. Customer-facing agents talk to callers, website visitors and senders; they are reception, support and first contact. Internal agents work for your team: rinqo's personal agent is a chat with your company knowledge, comparable to ChatGPT, except it knows your processes, prices and documents. A third axis is reactive versus proactive: reactive agents wait for incoming enquiries, proactive ones follow up, send appointment reminders and check in on open cases. In practice, SMEs combine several types: a phone agent for reception, a chat agent for the website, plus the internal assistant. At rinqo, all agent types share the same knowledge tree, so you maintain your knowledge once, not multiple times.

AI agents and German law: GDPR and the AI Act

AI agents can be deployed in Germany on a legally sound footing if two sets of rules are observed: the GDPR and the AI Act (Regulation (EU) 2024/1689). The GDPR leads to three practical obligations. First, a data processing agreement (DPA) under Art. 28 GDPR with the platform provider. Second, a clean hosting chain: rinqo hosts with Hetzner in Falkenstein and Nuremberg (ISO 27001:2022); language model processing runs contractually guaranteed within the EU and with no training on customer data. A purely European fallback provider is available as an opt-in failover. Third, transparency towards data subjects in your privacy policy. From the AI Act, Art. 50 is the key point for SMEs: from 02.08.2026, every person interacting with an AI system must be informed of that fact. rinqo activates this labelling by default; the wording is editable. Violations cost up to €15 million or 3% of global annual turnover. Art. 4 has already applied since 02.02.2025: anyone deploying AI must ensure sufficient AI literacy in the team, which includes brief documentation of usage limits and escalation rules. In short: the legal situation is not an obstacle but a checklist. With EU hosting, a DPA and active labelling, the core points are covered.

How much does an AI agent cost? Pricing and the maths

An AI agent runs at rinqo for a fixed monthly price, with all agent types included. The packages cover the range from sole traders to agencies, tiered by included credits. Credits are the usage unit for calls, chats and emails. If your allowance runs low, you can buy additional credits without changing packages. We handle the basic setup, with a clear monthly rate. For comparison, the alternatives: enterprise suites sit well above this price range and rely on sales-led processes with bespoke implementation. Building on open-source frameworks saves licence fees but costs developer time for build and operation, plus ongoing hosting and inference costs. For the business case, what counts in the end is how much routine the agents take off your team. How many requests they handle on their own, you can see in Statistics under “Handled without you”.