Chatbot information
AI chatbot information transforms scattered expertise into targeted, named knowledge building blocks. The clearer the name, instructions and content work together, the more helpful and comprehensible the X-ERP chatbot answers.
Chatbot information: Writing knowledge so that it can be found and understood
The central chatbot setting provides the framework; The chatbot information provides the individual knowledge modules. Each entry should reliably explain exactly one topic.
Administration > Initialization > AI Chatbot Information
Fields explained
- Name: unique title of the knowledge module. Use terms that users actually ask about.
- Instruction: describes when and how the AI should use the building block, for example “Explain the process step by step and state the necessary requirements”.
- Info: contains the approved specialist knowledge, relevant limits, exceptions and, if necessary, the correct reference to a further topic.
How to build good chatbot knowledge
- Create a clearly defined entry for each topic.
- Write the name in your users' language.
- Formulate the instruction in an action-oriented manner.
- Structure Info with short paragraphs, clear terms and verifiable statements.
- Test multiple wordings of the same question.
- Correct or remove outdated and conflicting entries.
Quality over quantity
A small body of well-maintained information is more helpful than a large, unverified collection of text. Document the person responsible internally and the test date so that the chatbot does not create false security tomorrow with the knowledge it learned yesterday.
Related topics
- For key users – understand and prepare processes › Basic settings › AI (artificial intelligence) › Chatbot settings
- For key users – understand and prepare processes › Basic settings › AI (artificial intelligence) › AI text info
- For key users – understand and prepare processes › Basic settings › AI (artificial intelligence) › AI email info
Frequently asked questions
What is the difference between chatbot settings and chatbot information?
The settings define the general framework; The information provides individual technical knowledge modules.
How large should a knowledge entry be?
As short as possible and as complete as necessary. An entry should cover a clearly defined topic without contradictory secondary content.
Why do different question wordings need to be tested?
Users ask the same thing in different ways. The test shows whether the knowledge component is reliably found and applied regardless of the wording.