Indexing
This module has the AI fill in metadata from the content of the document itself.
A question is asked per metadata ("does the document hold an insurance clause?") and the answer of the AI becomes the value of that metadata.
This module is used in the content import, archive import and synchronization flows.
Setting up the AI service is described on the Mistral integration page.
- Server access configuration: server access module that holds the location of the service and its authentication information, centralized for every application. It must hold the configuration of an AI server.
- A.I. model: AI model chosen in the list.
- Maximum tokens supported by this model: limit of the selected engine, retrieved automatically.
- Limit the number of tokens sent: caps the size of the text sent to the AI. A token stands for about 4 characters in French and 5 in English.
- Maximum tokens sent: limit in tokens (cannot exceed the maximum supported by the engine).
- Configure temperature: sets the variability of the AI answers: high, it diversifies them. Low, it makes them more predictable.
- Temperature: number between 0 and 10 (7 by default).
- A.I. language: language used to communicate with the engine, chosen in the list. With automatic detection, the engine works out the language of the document and uses the matching context and indexing questions.
- Context and instructions for A.I.: context passed to the AI. For instance: "The supplied document is a contract in English. Return answers in English only."
- Indexing answers: table of the metadata to fill in. Each row asks a precise question, in the context of the text excerpt. The answer becomes the value of the metadata.
- Metadata: metadata that receives the answer.
- Question: question defining the text to index. For instance: "Does the document hold a specific insurance clause? If so, return true, otherwise return false". "Does the document hold the name of the customer? If so, return the name of the customer, otherwise return an empty string".
The "Test model" button opens a window to submit a text and get the result live, with the model and the module as configured. The test relies on the questions defined in the module. If another question is typed, it only answers that one.
Flows
This module can be used in the following flows: