poc-realtime-ai-assistant  by disler

Realtime AI assistant POC using OpenAI's Realtime API

created 10 months ago
704 stars

Top 49.5% on sourcepulse

GitHubView on GitHub
Project Summary

This project provides a proof-of-concept for a real-time AI assistant named Ada, leveraging OpenAI's Realtime API. It targets developers and AI enthusiasts interested in building interactive, tool-using AI agents with capabilities like voice commands, web scraping, and database interaction. The primary benefit is a glimpse into the future of AI assistant engineering with real-time feedback and chained tool execution.

How It Works

Ada utilizes OpenAI's Realtime API to enable continuous, low-latency communication. The assistant chains tools, allowing it to perform multi-step actions like scraping web content, interacting with databases (SQLite, DuckDB, PostgreSQL), and generating diagrams. Personalization is achieved through a personalization.json file, controlling aspects like browser commands, assistant name, and SQL dialect. Memory management is handled via active_memory.json for persistent state.

Quick Start & Requirements

  • Install dependencies: uv sync
  • Run the assistant: uv run main or uv run main --prompts "Prompt 1|Prompt 2"
  • Prerequisites: OpenAI API Key, Firecrawl API Key.
  • Setup time: Estimated < 15 minutes.
  • Resources: Python 3.11+, uv package manager.
  • Links: OpenAI Realtime API, Firecrawl

Highlighted Details

  • Demonstrates chaining of tools for complex tasks.
  • Supports voice commands and CLI text prompts.
  • Includes modules for audio handling, database interaction, and memory management.
  • Allows customization of assistant behavior and toolset.

Maintenance & Community

This is a v0.3 proof-of-concept, described as buggy. No specific contributors, sponsorships, or community links (Discord/Slack) are mentioned in the README.

Licensing & Compatibility

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

Limitations & Caveats

The project is a v0.3 proof-of-concept and is described as buggy. It lacks interruption handling and transcript logging. Potential improvements include parallel tool execution and fixing audio clipping issues.

Health Check
Last commit

9 months ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
0
Star History
16 stars in the last 90 days

Explore Similar Projects

Feedback? Help us improve.