File Search

Semantic search across your documents. Upload files and let AI find and retrieve the exact information you need.

How It Works

1
📤

Upload

Upload files to a vector store

2
✂️

Chunk

Files are split into chunks

3
🧮

Embed

Chunks are embedded into vectors

4
🔍

Search

AI retrieves relevant chunks

Supported File Types

📄
PDF
.pdf
📝
Word
.docx
📊
Excel
.xlsx
🎨
PowerPoint
.pptx
📋
Text
.txt, .md
💻
Code
.py, .js, ...

Quick Start

Python
from mythicdot import MythicDot client = MythicDot() # Create a vector store vector_store = client.beta.vector_stores.create( name="Company Docs" ) # Upload files file = client.files.create( file=open("handbook.pdf", "rb"), purpose="assistants" ) # Add to vector store client.beta.vector_stores.files.create( vector_store_id=vector_store.id, file_id=file.id ) # Create assistant with file search assistant = client.beta.assistants.create( name="Doc Assistant", model="mythic-4", tools=[{"type": "file_search"}], tool_resources={ "file_search": {"vector_store_ids": [vector_store.id]} } )

Features

🎯 Semantic Understanding

Finds information based on meaning, not just keywords. Ask questions in natural language.

📚 Citations

Responses include citations back to specific passages in your documents.

⚡ Fast Retrieval

Vector search retrieves relevant chunks in milliseconds, even across thousands of files.

🔄 Automatic Updates

Re-upload files and the vector store automatically updates with new content.

💡 Best Practice: Chunk Size

For best results, use documents with clear structure (headings, paragraphs). The default chunk size works well for most documents, but you can customize it for specialized use cases.

Start Searching Your Files

Enable semantic search in your applications.

Vector Stores → Assistants API →