Skip to main content

Capabilities

URL Navigation​

This option allows the assistant to download content passed in the prompt from the internet. When this option is active You can indicate a web page in the format https://page.xx in the prompt text and the contents of the page will be passed to the assistant as context to offer a better response. It is possible to scroll through multiple pages, and they can all contribute to the context while increasing consumption.

Extract text from Web Pages​

This option allows you to choose whether the assistant should only read the contents of the web pages or treat them as html pages. In most cases it is sufficient to only read the contents of the page, optimizing the consumption of credits. However, if you need to be able to directly access the html code, deactivate this option to allow the assistant to directly consult the html structure of the web pages as well as the content.

This option allows you to enable the Web Search for the assistant. With the Web Search enabled, the user can enable the assistant to perform Web Searches autonomously in function of the topic being discussed.

Web Search Results​

You can configure the number of Web Search results to consider (max 10). More results can improve the quality of the assistant's answers but increase the number of tokens used by the same to formulate a response. The default value (5) is a good compromise between accuracy and costs.

Assistant Answer Mode​

There are 3 answer modes:

  • Open: The assistant tries to respond using the Knowledge Base but is allowed to use external information if it doesn't have an answer.
  • Closed: For each answer the assistant will rely exclusively on his Knowledge Base (RAG).
  • Hybrid: Combination of both approaches. if no data is found in the Knowledge Base (RAG), the assistant doesn’t respond but offer the chance to ask the same question in Open mode.

Response Length​

Sets the average length of the assistant's response.
The available values are:

  • Automatic
  • Very Synthetic
  • Synthetic
  • Normal
  • Long
  • Verbose

Attachments​

Configure the ability to send and receive attachments in conversations with the assistant.

Multi-Assistant Tagging​

Configure the ability to tag multiple assistants in conversations.

Capabilities Settings