Oliva is a multi-agent assistant framework designed for developers and researchers building AI applications that leverage semantic search, text generation, and question answering. It provides a flexible and extensible architecture for combining various agents into complex workflows, enabling sophisticated AI interactions.

How It Works

Oliva employs a modular, agentic RAG (Retrieval-Augmented Generation) architecture. It utilizes a graph-based workflow where agents process state through a series of nodes (functions) connected by conditional edges. This allows for dynamic routing and decision-making based on agent outputs. Key components include core agent interfaces, LangChain integrations for document retrieval and tool operations, and LiveKit integration for voice capabilities.

Quick Start & Requirements

• Install: uv sync to create a virtual environment and install dependencies.
• Prerequisites:
  • Qdrant Vector database (Docker recommended).
  • Deepgram account for speech-to-text.
  • LiveKit account for real-time voice.
  • Python 3.12+.
  • Environment variables (LIVEKIT_URL, LIVEKIT_API_KEY, LIVEKIT_API_SECRET, Qdrant API key) must be set.
• Setup: Follow instructions in tabular-semantic-search-tutorial or use assets/snapshot.zip for database population.
• Usage:
  • Voice assistant: make oliva-start
  • Agent playground: Download and run npm run start locally.
• Links: tabular-semantic-search-tutorial

Highlighted Details

• Integrates LangChain for LLM application framework.
• Leverages Qdrant for semantic search storage.
• Supports real-time voice communication via LiveKit.
• Utilizes Deepgram for speech-to-text.

Maintenance & Community

This project is open-source and welcomes community contributions. Further details can be found in the Contributing Guide.

Licensing & Compatibility

The README does not explicitly state the license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project requires multiple external service accounts (Deepgram, LiveKit) and a Qdrant instance, increasing setup complexity. The license is not specified, which may impact commercial adoption.