Knowledge Base

The knowledge base is the content your co-pilot draws from when answering questions. Connect your docs, upload files, or point it at your website — Pillar indexes everything and retrieves the right content at query time.

How It Works

Pillar uses retrieval-augmented generation (RAG) to ground the co-pilot's answers in your actual content:

1. You connect one or more content sources (website, files, cloud storage, or snippets)
2. Pillar indexes the content and generates embeddings for search
3. When a user asks a question, Pillar finds the most relevant sections
4. The co-pilot builds its response from those sections instead of relying on general AI knowledge

Sources

Mixing Source Types

Most setups use more than one source type:

• Website to index your public docs or help center automatically
• Files for internal guides, product manuals, or PDFs that aren't on the web
• Cloud Storage when you have a large document library managed in S3 or GCS
• Snippets to patch gaps, add FAQs, or give the co-pilot topic-specific instructions

You can add as many sources as you need. The co-pilot searches across all of them and pulls the best matches regardless of where the content lives.

Supported File Types

Supported file types differ by source:

File uploads:

Maximum file size: 100 MB

Cloud storage (S3 / GCS):

Next Steps

• Website — Crawl and index your public site
• Files — Upload documents directly
• Cloud Storage — Connect an S3 or GCS bucket
• Snippets — Add custom text content