Token
The smallest unit of text a language model reads and writes. A word often becomes one or more tokens.
Knowledge base
Context window.
How much text a language model can take into account at once, measured in tokens.
01Explanation
How it works
Everything the model should weigh up in an answer has to fit in the context window: the instructions, the parts of the material that were fetched and the conversation so far.
If the conversation gets very long or the material very large, not everything fits, and the system has to choose what goes in. That is one reason to fetch relevant passages instead of sending all the material every time.
02In practice
In customer service
For a customer in the chat, the context window rarely shows. It is, however, why the assistant remembers what was said earlier in the same conversation, and why material is kept clear and free of duplicates.
04Get started
See how it works for you
In a short demo we show how an AI assistant can answer your customers' questions, using your own material. It takes a quarter of an hour. From SEK 3,900 a month excl. VAT, with the exact price set after the demo.