owwn.ai / AI chatbot

AI chatbot, AI assistant or AI agent: which one suits your organisation?

The terms sound alike but describe different solutions. Here you will find the differences and how to recognise a good solution.

Updated November 2026 · Published by campfire

The difference

How the four solutions differ.

What matters is where the answers come from and what the solution does on its own.

Classic chatbot

Understands
Predefined questions and keywords
Answers come from
Fixed, stored texts
Does
Links to pages or forms
Risk
Fails to understand many questions

AI chatbot

Understands
Free questions in any wording
Answers come from
The general knowledge of the language model
Does
Answers questions
Risk
May invent answers that sound plausible

AI assistant

Understands
Free questions in over 100 languages
Answers come from
Only the sources you approve
Does
Answers questions with sources
Risk
Gaps where content is missing. The dashboard shows them.

AI agent

Understands
Questions and intentions
Answers come from
Your sources and connected systems
Does
Books appointments and hands over leads
Risk
Needs clear rules and approvals

owwn.ai combines the two highlighted solutions. The assistant answers only from your sources and completes tasks in your systems.

Why the source matters

A chatbot that makes things up costs trust.

Language models sound convincing even when they lack the knowledge. Anything that matters in advice therefore belongs in a controlled knowledge base.

2 in 3

bots tested in car retail confirmed a made-up vehicle model. Source: study by 9Senses AG

165'000+

questions answered by the assistant on jungfrau.ch since August 2025, solely from the website

0

successful attacks after more than a year in the Jungfraubahnen bug bounty programme

Checklist

How to recognise a good solution.

Use these eight points to compare providers. Ask each of them how they meet them.

  1. 01

    Answers only from your sources

    You decide which content and systems the chatbot can access.

  2. 02

    Source references

    Every answer shows where it comes from. Customers and your team can check it themselves.

  3. 03

    Tests before go-live

    Your specialists test the answers against clear criteria before customers see them.

  4. 04

    A clear answer when there is a gap

    If the chatbot finds no answer, it says so and puts the person in touch with your team.

  1. 05

    Analysis of open questions

    A dashboard shows which questions remain unanswered. They become new content.

  2. 06

    Data under your control

    You decide where the data is stored and which language model answers. You can export everything at any time.

  3. 07

    Security under test

    Regular tests check whether attackers can trick the chatbot into errors or extract data.

  4. 08

    Tasks done directly

    If needed, it books appointments or hands over leads straight into your systems.

In use

How owwn.ai assistants work.

Two examples from live operation. The case studies show the details and figures.

Tourism · Jungfraubahnen

Answers in 57 languages and sells tickets in the chat.

165'000+

questions answered since August 2025

−40 %

email workload at the customer centre

Tourism case study →

Automotive · Leapmotor Switzerland

Advises on models and hands over leads to the dealer.

44 %

of conversations show purchase intent

1 in 32

conversations ends as a lead with the dealer

Automotive case study →

How we ensure quality is explained in the white paper «The end of search».

Process

From prototype to go-live.

The assistant builds on your existing content. You see early on how it handles your questions.

  1. 01

    You supply what already exists

    Your content and access to your systems are enough. No new database is needed.

  2. 02

    Prototype after about five days

    With your real content and in your brand’s look. Your specialists test it first.

  3. 03

    Go-live after at least two months

    Depending on scope and integrations. After that, we support and develop the assistant continuously.

AI chatbot FAQ

Questions before you decide.

Book a demo
How much does an AI chatbot cost?

Projects start at CHF 16'000. Operation costs from CHF 100 per month, plus the cost of the language model. You will find details under «Process and pricing» on the home page.

Does an AI chatbot comply with Swiss data protection law?

Yes, if hosting and data processing are properly agreed. owwn.ai runs in Swiss data centres on request. We conclude a data processing agreement with you under the Swiss FADP.

Does an AI chatbot replace the customer centre?

No, it takes load off it. It handles the recurring questions. At Jungfraubahnen, the email workload at the customer centre fell by 40 %. Your team gains time for the requests that need a person.

Do we need a new database for this?

No. The assistant builds on your existing content and systems.

Does it also work on WhatsApp or by phone?

Yes. The same assistant answers questions on the website, on WhatsApp and by phone.

Contact

Talk to us about your assistant.

Book a demo