Projects

Projects are the top-level organizational unit in Opentrace. Each project is a self-contained workspace with its own documents, conversations, and retrieval settings.

What is a Project?

A project acts as a container that groups related documents together. When you ask the AI a question inside a project, it only searches through that project's documents — keeping your knowledge bases separate and focused.

Think of projects like separate research workspaces:

“Q4 Financial Report” — containing quarterly earnings, analyst reports, and board presentations
“Machine Learning Research” — with academic papers, textbook chapters, and lecture notes
“Product Documentation” — API docs, user guides, and release notes

Data Isolation

Each project is completely isolated:

Documents — uploaded files and URLs are scoped to the project
Conversations — chat history is per-project, so context never leaks between workspaces
Settings — RAG strategy, embedding model, and retrieval parameters are configured independently per project
Chunks & embeddings — vector data stays within the project boundary

Note

Deleting a project permanently removes all its documents, conversations, chunks, and settings. This action cascades through the database and cannot be undone.

Project Settings

Every project is created with default retrieval settings that you can customize at any time:

Setting	Default
RAG Strategy	Basic (vector search)
Agent Type	Simple
Embedding Model	text-embedding-3-large
Chunks per Search	5
Final Context Size	5
Similarity Threshold	0.3

See RAG Settings for a complete reference of all configurable parameters.

Was this page helpful?

PreviousSystem Requirements NextKnowledge Base