What does a knowledge graph cost? The honest answer
The short answer
It depends on your knowledge, not on the AI.
A system that checks invoices, writes quotes or answers customer questions costs exactly as much effort as the expertise it has to represent properly. Two businesses in the same trade, the same task, and two completely different figures at the end. What we can tell you up front: systems like this pay off above a certain recurring workload. If your business sits below it, we will tell you before you order anything.
Run the figures yourself
Our calculator shows what a system costs and when it starts to carry itself. It works with your hours and your hourly rate, with the formula printed underneath. And it tells you when it is not worth it for you yet.
Open the calculatorWhat actually drives the price
Seven things decide the effort. Answering questions is always included, everything else is added as needed. Worth noticing: none of the expensive items is technology.
- Depth of the knowledge base
- The biggest lever by far. One clear domain with fifty terms is a different thing from hundreds of entities with relationships between them.
- Photo and image recognition
- Recognising things in images is its own building block and adds a cost per request, because every evaluation takes computing time.
- Recommendation logic
- An answer turns into a decision. That needs rules somebody has to stand behind professionally.
- Multiple languages and country rules
- Different regulations and reporting duties in each country. That is not a translation problem, it is a rules problem.
- Acting on its own
- The system acts instead of only answering. Every action needs a limit and someone who draws it.
- Verified accuracy
- Having every line of knowledge checked by a domain expert means buying expert time. The most expensive and the most valuable part.
- Delivery
- A component for your own website is quick to build. A dedicated app is a second project alongside the first.
Why more than a chatbot?
An ordinary chatbot chats. A knowledge graph knows, because we curate, structure and secure your verified expertise so the AI only says what is actually true. That is handwork on knowledge, not a template. The payoff: your expertise becomes a product that recognises, recommends and works around the clock without anyone from your team being present.
Two businesses, two figures
Both wanted the same thing: an AI that answers from their own expertise. The effort ended up far apart, for reasons that had nothing to do with technology.
One had already ordered their knowledge
Checklists, rules and approvals existed in spreadsheets, maintained by a person who owns them. We had to structure, not reconstruct. The project was short accordingly.
The other had it in their heads
The knowledge sat in three heads, in old quotes and in emails. Before an AI could answer from it, someone had to write it down and someone else had to confirm it. That work is the price, not the technology.
That is why there is no list here. We look at how your knowledge exists today and work out the figure on that.
Running costs
What comes up monthly has three parts: running the knowledge base, maintaining the knowledge so it does not go stale, and the AI usage cost per request. The third part depends on how often the system is asked. We pass that on by actual usage instead of estimating a flat rate, so you do not pay for requests nobody makes.
Is a website with an assistant enough?
If you "only" need a website with a smart assistant, that is considerably cheaper and has a real fixed price, from €1,500. A knowledge graph is worth it once the AI has to recognise or recommend on genuine expertise.
What does an AI website cost? →Frequently asked questions about price
What does a knowledge graph or custom AI cost?
There is no list price, because the effort depends on the scope of the expertise: the number of rules, sources, approvals and interfaces. Two businesses with the same task end up at different figures, depending on how well ordered their knowledge already is. We look at it and work the figure out on that.
Why do you not name a price?
Because a number that applies equally to everyone would be wrong for a build project. Anyone publishing a list is selling a list product. What we disclose instead is the arithmetic behind it: when a system starts to carry itself, traceable in our calculator, with the formula printed underneath.
When is a system like this worth it at all?
Above a recurring workload that repeats week after week. The calculator at /werkzeuge/stunde-zurueck shows the threshold for your business, worked out with your hours and your hourly rate. If you sit below it, it says exactly that.
What drives the price the most?
Preparing the knowledge, not the model and not the hosting. If checklists, rules and approvals already exist in order, the project is short. If the knowledge sits in heads and old emails, writing it down and confirming it is the real work.
Why is this more than a chatbot?
Because a knowledge graph is bespoke: we curate and structure your verified expertise and secure it so the AI only says correct things. That is handwork, not a template, and it turns your knowledge into a product.
What is the difference to an AI website?
A website with an assistant has a fixed price from €1,500. A knowledge graph is a different thing: it recognises and decides on your verified expertise instead of only answering.
Put together your knowledge brain
Choose the scope. What it costs for you, we work out on that.
Open the knowledge brain