Giselle is an open-source AI agent studio designed for agentic workflows, facilitating seamless human-AI collaboration. It empowers both developers and non-engineers to build and deploy AI agents for various use cases, offering a visual interface and multi-model support to streamline complex processes and automate tasks.

How It Works

Giselle employs a visual, drag-and-drop interface for agent creation and modification. Its core strength lies in multi-model composition, enabling agents to dynamically select the most suitable AI provider (OpenAI, Anthropic, Google AI) for each specific task. A built-in knowledge store allows agents to access and search code and data, with support for GitHub vector store integration.

Quick Start & Requirements

To run locally: clone the repository, install dependencies using pnpm install, create a .env.local file with at least one AI provider API key (e.g., OPENAI_API_KEY="your_openai_api_key_here"), and start the development server with pnpm turbo dev. The application will be accessible at http://localhost:3000. A cloud-hosted version is also available.

Highlighted Details

• GitHub AI Operations: Automates GitHub issues, pull requests, and deployments using AI.
• Visual Agent Builder: Intuitive drag-and-drop interface for rapid agent design.
• Multi-Model Composition: Leverages various LLMs like GPT, Claude, and Gemini.
• Knowledge Store: Centralized access and search for code and data, with GitHub vector store integration.
• Team Collaboration & Template Hub: Features currently in development.

Maintenance & Community

The project is under active development. Contributions are encouraged via pull requests, bug reports, and feature requests. Links to social media (Facebook, X, Instagram, YouTube) are provided. The public roadmap is currently being finalized.

Licensing & Compatibility

Giselle is licensed under the Apache License Version 2.0.

Limitations & Caveats

The project is in active development, with features like Team Collaboration and the Template Hub marked as "In Development." The public roadmap is still being created, indicating potential shifts in direction or feature prioritization.