Index

Index

The Index section is where users create and manage indexes that organize their documents for efficient retrieval by AI agents. Indexes play a crucial role in enabling agents to quickly and accurately access the information required for specific tasks.

Creating a New Index

To create a new index, users should follow these steps:

1. Click the "Create Index" button.

2. Provide a name and description for the index to clearly define its purpose.

Choosing an Index Type

Users can choose between two types of indexes: Standard and Summary.

1. Standard Index:

   • A Standard Index is a vector-based index that organizes documents into manageable chunks, facilitating efficient search and retrieval by the agent.

   • Chunk Size: Users must select an appropriate chunk size based on the type of documents being indexed. For example:

     • Small Chunks: Ideal for documents such as user manuals or knowledge bases where detailed retrieval is essential.

     • Large Chunks: Suitable for longer, narrative documents like reports or articles where broader context is more beneficial.

   • After selecting the desired chunk size and choosing the documents to include, users can click Create Index to finalize the Standard Index.

2. Summary Index:

   • A Summary Index processes up to 5,000 characters from each document to generate a concise summary.

   • Summary Query: Users must provide a summary query and select a suitable Large Language Model (LLM) to execute the query.

   • The generated summary is then stored in the vector index, enabling quick retrieval during agent interactions.

   • Once the summary is created, users can click Create Index to complete the setup of the Summary Index.