Knowledge Spaces Guide - File-Based AI Context in Rephlo

What are Knowledge Spaces in Rephlo?

Knowledge Spaces are persistent document containers that give AI access to your files. Upload PDFs, Word documents, code files, or text, and Rephlo includes their content as context when processing your requests. This produces grounded, accurate answers based on your actual documents.

Think of a Space as a folder of reference materials. Without a Space, the AI relies on general training data. With a Space active, the AI uses your uploaded files to generate responses. For example, you can upload a project specification and ask the AI to "draft a release email based on the uploaded ProjectSpecs.pdf". Rephlo supports .pdf, .docx, .txt, .md, and code files.

How do I create and use a Knowledge Space?

Open the Dashboard, click New Space, give it a name, and drag files into it. Rephlo extracts and stores the text content. When you run a command, activate the Space to include its documents as AI context. You can switch between Spaces instantly.

Rephlo uses Full Context Injection — all Space content is included in the prompt sent to the AI. This means the AI reads everything in your Space, not just relevant snippets. Token budgets are managed automatically, and Smart Compaction summarizes older documents when a Space gets too large.

1. Create a Space: Open the Rephlo Dashboard, click "New Space", and give it a descriptive name (e.g., "Q3 Analysis" or "Project Docs").
2. Add files: Drag and drop files into the Space, or click "Add Files" to browse. Supported: PDF, DOCX, TXT, MD, and code files.
3. Activate the Space: Select the Space from the Dashboard or Command Runner dropdown to make it active for your next interaction.
4. Run a command: Select text and run a command as usual. Rephlo automatically includes all Space documents as context for the AI.

What file types do Knowledge Spaces support?

Knowledge Spaces support PDF (.pdf), Word (.docx), plain text (.txt), Markdown (.md), and code files including .cs, .py, .js, .json, and .log files. File content is extracted as plain text for AI processing. Binary files and images are not supported.

For best results, use text-rich documents. Large PDFs are processed page by page. Code files retain their formatting for accurate code analysis. If a Space exceeds the AI model's context window, Rephlo's Smart Compaction feature automatically summarizes less critical content to stay within limits.

Frequently Asked Questions