Blog · Published · Updated
How to compare AI chatbots for a business
Short answer
You compare AI chatbots on real customer questions and approved material. Fluent answers are not enough. The chatbot must keep the terms from the source, admit when information is missing, follow a change in the conversation and hand the case to an employee with its context. At Syntalith the app costs from €3,000 net, and with 1,500 customer questions a month at 4 minutes each it pays back in 1.2 months.
How we workDetails ↓
- 01Call30 minutes with an engineer, free
- 02PrototypeA working system on your process and data, free
- 03ImplementationTarget and fixed price in the contract; 50% at signing
- 04Acceptance50% once the agreed target is met
- analysis with an engineer and a free prototype
- €0
- team in ChatGPT or Claude, up to 10 people
- from €1,150
- automation, agent or application for one process
- from €3,000
We belong to the OpenAI and Anthropic partner programmes and learn through their official courses. What this means ↓
Author: Syntalith · 3 min read
Statistics
Chatbots and customer service: statistics 2026
1.2 mo.
payback of a €3,000 app at 1,500 customer questions a month: €3,490 of work today, €1,059 with checking and the model. Scenario.
15%
According to Brynjolfsson, Li and Raymond (2025), customer service agents supported by AI resolved on average 15% more issues an hour, in a study of 5,172 people.
Source: Brynjolfsson, Li and Raymond, The Quarterly Journal of Economics (2025) ↗
30%
According to Brynjolfsson, Li and Raymond (2025), less experienced agents with AI raised the number of issues resolved an hour by 30%.
Source: Brynjolfsson, Li and Raymond, The Quarterly Journal of Economics (2025) ↗
47%
According to the European Commission (2026), customer service is the most common use of AI among EU SMEs and mid-caps that already use it, at 47%.
Source: European Commission, Flash Eurobarometer 578 (2026) ↗
2 August 2026
According to the European Commission (2026), the transparency obligations in Article 50 of the AI Act apply and are enforced from 2 August 2026, and providers of systems placed on the market earlier have until 2 December 2026 to mark content under Article 50(2).
Figures checked: 22 October 2026
Questions to prepare before you watch a demo
Choose an ordinary question about the service, a question that needs clarifying and a case the documents do not answer. For each one, write down the expected scope of the answer and the source. In the example, the current service description says the date is agreed individually. A chatbot that answers „done tomorrow” invents a commitment. The correct answer says that a consultant sets the date and offers contact.
After the first answer, change a detail of the case. The customer may talk about a different product from the one at the start, or correct the quantity. Check whether the chatbot carries the terms over to the new product or repeats the old ones. If a shop's chatbot can see orders, repeat the trial after logging out and switching accounts: data fetched earlier must not stay in the conversation.
| Area of the trial | Example of an error |
|---|---|
| Answer from the source | an added price or date |
| Missing information | a confident answer with no basis |
| Change of context | a condition that concerns the previous product |
| Handover | an important detail missing for the employee |
- 01Questions30 to 50 customer questions with the expected scope of the answer
- 02Conversationsa change of product, quantity and account during the conversation
- 03No basisquestions the documents do not answer
- 04A persona request for a consultant during working hours and after them
- 05Assessmentcontent and handover scored separately
- The employee receives the question and what the conversation established
- The customer knows the case is waiting and roughly when to expect an answer
- After working hours the chatbot does not promise immediate contact
- The notice that the customer is talking to AI appears at the first contact
What an AI chatbot costs and when it pays back
- Today
- 1,500 customer questions × 4 min = 100 h × €34.90 = €3,490
- With the system, a person checks the output in 30% of the time
- €1,047
- Model, 1 call per case, Claude Sonnet 5.5
- €12
- Cost of one case
- €2.33 today, €0.71 with the system
- Saving
- €2,431 a month
Payback of €3,000
1.2 months
The volume, the time and the checking share are the scenario's assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat. Model: 3,000 input tokens and 300 output tokens per call at Anthropic's rates, converted at the ECB rate of 1.1205 USD per euro on 22 October 2026.
The count covers the work around a case: searching, reading, writing and passing it on. What stays is the time to check the output, 30% of today's work in the scenario. Before we quote, we measure both times on 20 to 100 cases your team has already handled, in a €0 test that takes 5 working days.
The cost of an answer includes the corrections. In the count we leave 30% of today's work for cases passed to a consultant and for checking. According to Brynjolfsson, Li and Raymond (2025), customer service agents supported by AI resolved on average 15% more issues an hour (QJE). So it is worth comparing a chatbot and a consultant with an assistant on the same set of questions.
Content and integration, tested separately
You can start the content test on made-up material. The connection to the ticketing system needs its own trial, including an interruption and a retry. Recording tickets after hours is described in the AI chatbot for customer service use case. You check the notice about talking to AI at the first contact. The obligations in Article 50 of the AI Act apply from 2 August 2026, and the details are in the article on AI Act Article 50 and chatbots.
Repetitive questions with answers in the documents
a chatbot that cites the source and refuses when there is no basis
from €3,000
Many cases need a consultant
a chatbot that hands over the case with what it established
from €3,000
Answers depend on customer data in the system
a chatbot with access to orders and account checks
priced after the review
Where to start
The first test covers 30 to 50 questions from the last month and one path for handing over to a consultant. We also record the questions the chatbot should not answer on its own. We build this kind of rollout as part of AI apps. For the free process review, a description of one case from recent weeks and the tools it went through is enough. You do not need to send confidential documents, and you get a written plan within 2 working days.
Worked example · Scenario
How we count the payback: an AI chatbot for a business
We run the same count for your customer questions in the free process review.
€2,327
a month for answering 1,000 customer questions
Payback: 1.5 mo.
4 minutes for an employee to answer a question and checking in 15% of the old time are our assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat, for companies with 10 or more employees. Model cost: Claude Sonnet 5.5, about 4,000 + 400 tokens per answer, at the ECB rate of 1.1205 USD per euro on 22 October 2026.
See the full calculation →
