explore
swipe
Please turn off low power mode for a smoother experience.
Artificial Intelligence

An assistant that knows your answers

We build chatbots and AI assistants that work with your own documents and product data. They answer questions with what is actually written down at your company, not with what sounds generally plausible.

Let's connect 🚀

What this is about

A chatbot is only as good as the knowledge it can reach. The difference between a useful assistant and an embarrassing one is not the language model, but where the answers come from and what happens when it does not know something.

Answers from your sources

The assistant draws on manuals, data sheets, price lists or the ticket system. Anything not covered there is not invented, it is passed on as an open question.

Limits are part of the solution

We define what the assistant gives information about and where it hands over to a person. With prices, commitments and legal matters, that boundary counts for more than any feature.

Not every question needs a bot

When the same five questions keep coming in, a better page often helps more than a chat window. We say so when that is the more honest route.

A bot that guesses costs more trust than it saves.

Situations that typically bring clients to us

What we actually do

Assistants for customers

A chat on your website or in your shop that answers questions about products, delivery times and processes from your own data.

Assistants for staff

Internal search across manuals, policies and project knowledge. Particularly useful where hard-won experience sits with individual people.

Connection to your systems

A link to your shop, ERP, PIM or ticket system so the assistant works with current data instead of an outdated snapshot.

Handover to people

Clear rules for when a case goes to the team, including a summary of the conversation so far.

Data protection and operation

We clarify which model runs where and which data it is allowed to see. In an environment inside the EU if you prefer.

Control in operation

We look at the questions people actually ask and improve the knowledge base wherever answers fall short.

A matching journey

Screenshot RHEWUM

RHEWUM AI Cockpit

A central AI platform with secure access for everyone in the company, instead of many isolated tools scattered across departments.

View journey

How we work

  • 01

    Collect the questions

    First we look at what customers or colleagues genuinely ask. That decides whether an assistant is worth it, and what for.

  • 02

    Organise the knowledge

    Gather the sources, sort out duplicates and outdated material. This step determines the quality of every answer that follows.

  • 03

    Build and test

    Set the assistant up and test it with real questions before it is let loose on customers.

  • 04

    Support

    Watch the answers in operation, close the gaps and widen the scope step by step.

The effort is not in the bot, it is in the order behind it. Send us your ten most common questions and we will tell you whether an assistant is worth it.

Let's connect 🚀

More journeys

Two projects where AI does everyday work: once as a central access point for all staff, once directly inside the editorial system.

RHEWUM AI Cockpit project

RHEWUM AI Cockpit

A central AI platform with secure access for everyone in the company, instead of many isolated tools scattered across departments.

View journey
Deutsches Handwerksblatt project

Deutsches Handwerksblatt

AI in the daily editorial routine, with suggestions that follow the publisher's guidelines instead of generic boilerplate.

View journey

Related services

An assistant is rarely the first step. Automation, a clear AI strategy and connected systems usually belong to it.

Frequently asked questions

A bot that guesses costs trust. It pays off when the same questions can be answered from your own documents.

When is an AI chatbot worth it?

When similar questions arrive every day that can be answered from existing documents, or when knowledge sits in PDFs that nobody can search through. The benefit is just as large internally, for example when new team members take a long time before they can answer questions, or when hard-won experience sits with individual people. Another typical case is enquiries outside business hours that would otherwise be left unanswered.

Does the assistant work with our data?

Yes, and that is the decisive point. The basis is your manuals, data sheets, price lists or tickets, not general internet knowledge. On request we connect your shop, ERP, PIM or ticket system so the assistant works with the current state rather than an export from a year ago. Anything not covered in your sources is not invented, it is passed on as an open question.

Is a better FAQ page sometimes enough?

Yes, and we will say so. If five recurring questions carry most of the load, a well-written page is cheaper, faster and valuable for search engines on top. A chat window pays off when the questions are varied, when follow-ups come up or when answers depend on the product, the order or the person asking.

How do you deal with wrong answers?

On three levels. Clear limits on what the assistant gives information about at all: with prices, commitments and legal questions it hands over to a person, along with a summary of the conversation. Tests with real questions from support before going live. And ongoing control in operation: we look at what people actually ask and improve the knowledge base wherever answers fall short.

The same questions in support over and over?

Send us your ten most common questions. We will tell you whether an assistant helps or a better page does more.

Let's connect 🚀