Skip to content

🤖 AI-Powered Features

Redoc leverages AI to enhance document processing capabilities. This guide covers the AI features and how to use them effectively.

AI Capabilities

1. Smart Document Processing

  • Content Analysis: Extract key information from documents
  • Document Classification: Automatically categorize documents
  • Entity Recognition: Identify and extract entities (names, dates, amounts, etc.)
  • Sentiment Analysis: Analyze the sentiment of text content

2. Content Generation

  • Text Summarization: Generate concise summaries of documents
  • Content Expansion: Expand bullet points into full paragraphs
  • Document Outlining: Create structured outlines from unstructured text
  • Language Translation: Translate content between languages

3. Data Extraction

  • Form Recognition: Extract data from forms and invoices
  • Table Extraction: Convert tables in documents to structured data
  • Key-Value Pair Extraction: Extract structured data from unstructured text

Getting Started with AI Features

Prerequisites

  1. Install Redoc with AI support: 
    pip install "redoc[ai]"
  2. Set up your AI provider (OpenAI, Ollama, etc.)

Basic Usage

Text Summarization

from redoc import Redoc

redoc = Redoc()
summary = redoc.ai.summarize("long_document.txt", length="brief")
print(summary)

Entity Recognition

entities = redoc.ai.extract_entities("document.pdf")
for entity in entities:
    print(f"{entity['text']} - {entity['type']}")

AI Providers

Redoc supports multiple AI providers:

1. Ollama (Local)

from redoc.ai.providers import OllamaProvider

provider = OllamaProvider(model="mistral:7b")
redoc = Redoc(ai_provider=provider)

2. OpenAI

from redoc.ai.providers import OpenAIProvider

provider = OpenAIProvider(api_key="your-api-key", model="gpt-4")
redoc = Redoc(ai_provider=provider)

3. Custom Provider

Implement your own provider by extending the AIProvider class:

from redoc.ai.base import AIProvider

class MyAIProvider(AIProvider):
    def summarize(self, text: str, **kwargs) -> str:
        # Your implementation here
        return summary

Advanced Usage

Document Q&A

answer = redoc.ai.ask("document.pdf", "What is the main topic?")
print(answer)

Document Comparison

diff = redoc.ai.compare_documents("doc1.pdf", "doc2.pdf")
print(diff)

Best Practices

  1. Data Privacy: Be mindful of sensitive information when using cloud-based AI services
  2. Cost Management: Monitor API usage to avoid unexpected costs
  3. Quality Control: Always review AI-generated content
  4. Performance: Cache results when processing multiple documents
  5. Fallbacks: Implement fallback behavior for when AI services are unavailable

Configuration

Configure AI settings in ~/.config/redoc/config.yaml:

ai:
  provider: ollama  # or 'openai', 'custom'
  model: mistral:7b
  temperature: 0.7
  max_tokens: 1000

  # Ollama specific
  ollama:
    base_url: http://localhost:11434

  # OpenAI specific
  openai:
    api_key: ${OPENAI_API_KEY}
    organization: ${OPENAI_ORG}

Troubleshooting

Common Issues

  1. Missing API Key 

    export OPENAI_API_KEY='your-api-key'

  2. Model Not Found 

    # For Ollama
ollama pull mistral:7b

  3. Rate Limiting

  4. Implement retry logic
  5. Reduce request frequency
  6. Check your API quota

Next Steps