TeamAI's Knowledge feature, also known as Datastores, allows builders to provide additional context to their AI assistants (GPTs) by uploading relevant files. This article will guide you through the process of using Knowledge in your assistants, highlighting its benefits and best practices.
What is Knowledge?
Knowledge, or Datastores, is a feature in TeamAI that enables builders to attach multiple collections, datastores, documents, and files to an assistant. Unlike ChatGPT, which only allows attaching up to 20 files to a GPT, TeamAI provides greater flexibility in the number of files you can include.
How does Knowledge work?
Using the assistant configure editor tab, select the desired collections, datastores, or even documents to your assistant.
The GPT breaks the text into chunks, creates embeddings (mathematical representations of text), and stores them for later use.
When a user interacts with your GPT, the assistant can access the uploaded files to obtain additional context and augment the user's query.
The GPT chooses one of the following methods based on the requirements of the user's prompt:
Semantic search: Returns relevant text chunks, preferred for "Q&A" style prompts where a specific portion of the source document is required.
Document review: Returns entire short documents and/or relevant excerpts of larger documents, including them along with the prompt as additional context.
The Knowledge Access Process
You ask your assistant a question related to the information contained in your datastores.
The assistant searches through the selected datastores to find relevant information.
The assistant uses this information to provide a more accurate and contextually relevant response to your query.
By granting your assistant access to datastores, you can unlock their full potential and enable them to provide more valuable insights and assistance.
When to use Knowledge
Currently, the only way to manage files attached to a GPT in TeamAI is through the Datastores (Knowledge) feature. This makes it best suited for applications where context changes infrequently, such as employee handbooks, policy documents, or school curricula.
For more information on Datastores, refer to the following related documents:
Tips for getting the most out of Knowledge
- Assistant to rely on Knowledge first before searching the internet or using any other custom plugin.
- Assistants to "cite their sources" when providing information from the datastores.
Best Practices for Datastore Access
Only select collections and datastores that are relevant to the tasks your assistant will be performing to keep their knowledge focused and improve response accuracy.
Regularly review and update the datastores your assistant has access to, ensuring they have the most up-to-date information.
If you have sensitive information in your datastores, consider creating separate collections or datastores specifically for your assistant's use, rather than granting them access to your entire library.
By leveraging the power of Knowledge in TeamAI, you can create AI assistants that provide more accurate, context-aware, and valuable responses to user queries.