Understanding Knowledge Base

About

A Knowledge Base is a store of your organization’s content that Future AGI indexes and makes available across the platform. When you upload documents, the platform processes and indexes them so they can be used as grounding context for synthetic data generation and evaluations.

When to use

• Synthetic data generation: You want generated examples to reflect your domain, terminology, and procedures instead of generic text. Selecting a KB when creating a synthetic dataset grounds the output in your actual content.
• Evaluation: You want to check whether model outputs are factually consistent with your organization’s knowledge. The KB supplies the reference context that hallucination detection and grounding evals compare against.

Supported Content Types

You can upload the following file types to a Knowledge Base:

File Type	Extensions
Word documents	.doc, .docx
PDF documents	.pdf
Plain text	.txt
Rich text	.rtf

Maximum file size is 5MB per file.

Examples of content that works well in a KB:

• Technical documentation and manuals
• FAQs and troubleshooting guides
• SOPs and process workflows
• Training materials and HR policies
• Legal documents and compliance information
• Product descriptions and specifications

File Processing

After you upload files, the platform processes them automatically. Each file goes through one of three states:

Status	Description
Successful	Content extracted and indexed for use
Processing	File is being processed
Failed	Processing failed. You’ll be notified and the file won’t be usable.

Next Steps

• Create KB Using UI: Upload files through the dashboard
• Create KB Using SDK: Upload and manage knowledge bases programmatically
• Synthetic Data: Learn how KB grounds synthetic data generation