Set up an FAQ chatbot: answers from your own knowledge, around the clock
You set up an FAQ chatbot by giving an AI access only to your approved content, such as FAQs, product information and price lists, letting it answer standard questions from that content with a source, and handing over to a person as soon as an answer is missing or the request is individual. Unanswered questions reveal gaps in your content. In the worked example with 40 requests a day this saves about 26 h a month.
- Time saved in the worked example
- approx. 26 h per month
- Delivery
- 7 to 12 working days
- Typical systems
- Website, Customer portal, WhatsApp Business
- Fixed price
- from €990
, Rouven Jahnke
Flow: Set up an FAQ chatbot
5 steps-
Question asked
On the website, customer portal or messenger
-
Look up knowledge
Only in your approved content
-
Answer
With source, labelled as AI
-
Hand over to a person
When no answer exists or it gets sensitive
-
Close gaps
Unanswered questions become content tasks
Green runs automatically. Orange is decided by a person.
Where time gets lost today
- The same questions over and over
- Customers wait until the next working day
- Answers vary depending on who replies
A large share of customer requests is routine: when do you deliver, what does it cost, how do I cancel, where is my invoice? Each answer is quick, but together they absorb time the difficult cases need.
Classic decision-tree chatbots frustrate customers as soon as a question is phrased slightly differently. And an unconstrained AI bot will make up answers when in doubt. Both damage trust.
How the automation works
The chatbot answers only from content you approve: FAQs, product descriptions, price lists, delivery terms, guides. For each question it first retrieves the matching section, and the AI phrases the answer only from that, with a reference to the source.
If no matching information exists, the bot says so honestly and offers a hand-over: as a ticket with the conversation so far, a callback request or directly to a person in chat. For certain topics, such as complaints or contract questions, it always hands over.
Questions the bot couldn’t answer are collected. You see what customers are looking for and what is missing from your content.
What your team still decides
People handle everything beyond standard questions: complaints, contracts, special cases, goodwill. Your team maintains the knowledge base; the bot only says what is in it. Users can always tell they are talking to an AI.
Calculate your savings
Adjust the values to your company. The calculation runs only in your browser; nothing you enter is transmitted.
- Working time saved per month
- – h
- Savings per month
- –
- Savings per year
- –
Excludes running costs for hosting and AI usage. You receive the implementation cost as a fixed price after the intro call.
Formula
Hours saved/month = Inquiries/day x 22 x Resolution rate x Handling time x Automation rate / 60
Worked example
The calculator above starts with these assumptions. They are example values, not averages from studies.
- Inquiries per day
- 40
- Auto-resolution rate
- 60 %
- Avg. handling time per inquiry
- 4 min
- Share of work automated
- 75 %
- Labour cost per hour (fully loaded)
- 38 €
Result: about 26 h less manual work per month, roughly €1,000 a month or €12,000 a year.
Before and after
| Aspect | Today, by hand | With the workflow |
|---|---|---|
| Availability | Office hours | Around the clock for standard questions |
| Answer quality | Varies by person | Consistent, with a source |
| Difficult cases | Wait in the same queue | Get more attention |
| Knowledge gaps | Invisible | Visible as a list of open questions |
Frequently asked questions
What happens when the bot doesn’t know something?
It says so openly and hands over to a person, for example as a ticket with the conversation so far. The bot is built to answer only from your approved content and not to invent facts.
Does the chatbot have to be labelled as AI?
Yes. Under the EU AI Act, people must be able to tell that they are interacting with an AI system. The bot therefore introduces itself clearly as a digital assistant and always offers a hand-over to a person.
How is the bot kept up to date?
It reads your content directly, for example in Notion, Confluence, on your website or in documents. When you change something there, the bot answers with the new information from then on. No retraining is needed.
What does this workflow cost and how quickly is it live?
Implementation is a fixed price from €990; the exact price is set after the free intro call and depends on the number of connected systems and exceptions. Typical delivery time for this workflow: 7 to 12 working days. Running costs for hosting and, where used, AI usage are quantified in the proposal.
Set up an FAQ chatbot: free intro call
Describe a process that eats time today. I’ll tell you honestly whether automating it pays off, what it would look like and what it costs.
- We walk through your process together, ideally on screen.
- You learn which step has the biggest leverage and what it takes.
- You receive a fixed-price proposal within 48 hours. No obligation.
Pick a time in the calendar
Loading calendar…
Thanks, your message arrived.
You’ll get a confirmation email. I’ll reply within 24 hours.
Your reference number: