How much does an AI chatbot cost for a business?
It is the first question we get, and the hardest to answer with a single number. An AI chatbot can cost anywhere from a few hundred euros a month to a project running into five figures. The difference is not how “good” the chatbot is — it is what you are asking it to do.
Here is what you are actually paying for, so you can compare quotes without guessing.
The four things that set the price
1. How much knowledge it has to learn
A chatbot answering 20 common questions is a completely different project from one that knows 8,000 products, their prices, stock levels and your returns policy.
The cost here is not the technology. It is the work of gathering your material, cleaning it up and giving it to the chatbot in a form it can use. If your material is already organised, you save money. If it is scattered across PDFs, emails and two people’s heads, it will cost more.
2. Whether it needs to reach your systems
A chatbot that says “returns are accepted within 14 days” is reading a document. A chatbot that says “your order shipped yesterday and arrives tomorrow” has to talk to your e-shop and your courier.
Every connection to an ERP, CRM, e-shop or shipping system adds cost. It is also where the chatbot starts saving real time, because those are exactly the questions your team is most tired of answering.
3. How many channels it runs on
Just your website? Or Messenger, WhatsApp and Viber too? Each channel has its own rules and its own work. Start where your customers already write to you, not everywhere at once.
4. Who maintains it
This is the cost most people forget. Prices change, products change, policies change. A chatbot nobody updates starts giving wrong answers within months — and at that point it does more damage than the value it was adding.
Always ask: who updates it, how often, and what does that cost?
What to watch for in a quote
“Unlimited conversations” — usually is not. Ask to see what happens when you go past the limit.
The AI cost, separately. Most chatbots run on a model that charges per use. Ask whether that is included or arrives as a separate bill. At aiEra you can use your own key, so you see exactly what you are paying.
Who owns the data. If you change provider next year, do you take the knowledge you built with you, or start over?
Greek. Do not assume it. Ask for a test with real questions from your own customers, including spelling mistakes and greeklish, because that is how people actually write.
When it is worth it, and when it is not
A chatbot pays off when you get the same questions over and over. If your team answers “how much is shipping”, “do you have the X in medium” or “when is my order arriving” every day, the maths is simple: how many hours go there each month, and what do those hours cost?
It does not pay off when every question is different and needs judgement. In that case the chatbot hands almost everything to a human anyway, and you have paid for an extra step.
It is also the wrong investment if your real problem is somewhere else. If customers keep asking where their order is because they never get an update, fix the updates first.
How to start without risk
Do not start with the perfect chatbot. Start with the ten questions you hear most, on one channel, with a clean handover to a person when it does not know.
You will have something live within weeks, you will see real numbers, and then you decide where it is worth expanding. That is far cheaper than building for six months and discovering nobody uses it the way you expected.
If you want a realistic estimate for your own situation, book a free 30-minute call. We will also tell you if we think it is not worth it for you.