Document Agent (Anthropic + Ollama Embeddings)

A Python starter project for a document assistant that can:

• read a PDF
• generate a summary with Claude
• suggest questions with Claude
• answer user questions with Claude using retrieved supporting chunks
• generate local embeddings with Ollama

Setup

1. Create a virtual environment.
2. Install dependencies:

   pip install -r requirements.txt

3. Add a .env file in the project root:

   ANTHROPIC_API_KEY=your_key_here
   ANTHROPIC_MODEL=claude-3-5-haiku-latest
   OLLAMA_BASE_URL=http://localhost:11434
   OLLAMA_EMBED_MODEL=mxbai-embed-large

4. Make sure Ollama is running and the embedding model is available:

   ollama pull mxbai-embed-large

Run summary mode

python main.py summarize path/to/file.pdf

Run question-answer mode

python main.py ask path/to/file.pdf "What are the key risks discussed?"

Suggested questions

python main.py suggest path/to/file.pdf

Notes

• V1 supports PDF only.
• FAISS is used as the local vector index.
• Traces are saved in runs/.
• Claude is used for summary and question answering via the Messages API.
• Ollama /api/embed is used for local embedding generation.