A chatbot or voice agent that answers client questions, even after 5 PM

Your client service gets the same questions every day. Where is my order, what are your opening hours, can I move my appointment. During the day, your team answers them between the real work. After 5 PM, the phone goes unanswered and the inbox piles up. A chatbot or voice agent takes over the first part of that contact, answers right away from your own data, and hands over to a colleague as soon as it gets complicated.

What a chatbot should and should not do

Most chatbots you meet are a choice menu with a bit of chat around it. They recognize five questions and send the rest to a contact form. That frustrates the client and saves your team nothing. We build an assistant on a language model that understands a question in plain language, even when it is worded differently from what you expected, and answers from your sources: your product data, your terms, the status of an order in your ERP.

What it does not do matters just as much. It does not invent a delivery date that is not in the system. If a client is angry, has a complaint, or asks something out of its reach, it passes the conversation to a staff member with the full context. No loop of "I don't understand your question". A proper handover.

Answers from your own knowledge, not the internet

A language model knows a lot, but nothing about your company. We load your manuals, frequently asked questions, price lists, and terms. For every question, the model looks up the relevant passages there before it answers. If the client wants to know where a package is, the assistant looks it up live in your online store or ERP through an integration, like the ones we build in System integration. We use the same technique in AI document search, only there it serves your own employees instead of your clients.

The model runs on a server inside the EU. Conversations with your clients do not go to a public service outside Europe and are not used to train anyone else's models. Personal data in a chat falls under the GDPR, so we record what is stored and for how long, and we arrange the matching documentation through GDPR compliance.

Calling a voice agent

Some clients do not chat. They call. A voice agent picks up the phone, understands Dutch, and holds the same conversation as the chatbot, only out loud. Moving an appointment, reporting a fault, or asking for an order status can happen at 11 PM without anyone on your team being awake. We make it say clearly that it is a computer, and with one sentence the caller reaches a person if they want one.

For a healthcare organization or a municipality, that means a shorter queue without more people on the phone. For a wholesaler, it means a technician on site can dictate an order instead of typing an email in the evening.

On the line in 10 business days

We start with a free intake in which we go through your current client contact together: which questions come in most, which can be taken out, and where the assistant must never decide on its own. Then, within 10 business days, we build a working prototype on your own sources. Your team tests it with real questions from the mailbox and the phone log. What the assistant gets wrong shows up in the overview, and we fix it. Only when the answers are right do we put it on your website or on the line.

You own the configuration, the knowledge base, and all conversation logs. If, over time, you want the assistant to carry out actions itself, such as creating a return or looking up an invoice, we extend it into one of our AI agents for the back office. And if the intake shows that a better FAQ page or a good email template would help you more, we tell you right away.