Description:
Allow users to specify which parts of the document or schema to extract, based on a custom prompt.
Tasks:
- Implement user-defined prompts (e.g., "extract only financial-related entities").
- LLM should filter the output to match the user’s prompt and context.