Giselle: A Visual AI App Builder Now Open-Source for the World — Enabling Anyone to Intuitively Create AI Applications

ROUTE06, Inc. (Headquarters: Chiyoda-ku, Tokyo; CEO: Takafumi Endo) is pleased to announce the official worldwide release of "Giselle," an open-source platform that enables anyone to intuitively build AI applications. Giselle is a platform that allows users to build and run AI applications through intuitive drag-and-drop operations without any programming required. It is designed as an execution infrastructure for "completing complex, long-running AI tasks as business operations." Without technical knowledge, all professionals—including engineers, designers, product managers, marketers, sales representatives, and back-office staff—can create AI applications tailored to their work and start using them immediately. Giselle is also released as open source under the Apache License 2.0, allowing individuals and companies worldwide to freely use, modify, and redistribute it.

Service Website: https://giselles.ai

Making Ever-Evolving AI Accessible to Everyone

AI is evolving at an extremely rapid pace, and it has become difficult for many people to keep up with and master the specifications of the latest models offered by various AI model providers.

Chat-based interactions with AI through services like ChatGPT, Claude, and Google Gemini have already become familiar to many people. However, combining multiple AI models to build workflows and custom AI applications still presents significant barriers.

While numerous AI tools exist today, most require engineering knowledge to use effectively. On the other hand, for engineers, it's often faster to call APIs directly and build solutions themselves rather than use such tools—leaving a gap for the right tool that doesn't quite exist.

Within our company, we heard requests for "an AI tool that non-engineers can use intuitively," so we decided to build it ourselves. That's how Giselle was born. We created a product we wanted to use ourselves and actually use it in our daily work.

Giselle's design philosophy is to reconstruct the AI orchestration that engineers were building themselves into a form that anyone can use. Even AI workflows that run for several hours can be safely executed while tracking progress. By separating construction and execution, we've achieved an experience designed with the premise of "completing" tasks as business operations.

We aim to democratize AI and create a world where anyone can intuitively build, use, and improve AI applications.

Key Features of Giselle

1. Visual Node Editor

Design AI workflows simply by dragging, dropping, and connecting nodes—like drawing a flowchart. Even without programming experience, you can visually construct complex AI processes.

2. Comfortable Execution of Long-Running Tasks

Created AI applications can be immediately used in a dedicated execution environment separate from the construction environment. You can monitor the progress of long-running AI applications in real-time, and multiple outputs and lengthy results are displayed in an easy-to-read format. By clearly separating construction and execution environments, daily AI application use becomes comfortable.

3. Multi-Model AI Collaboration

Freely combine world-leading AI models such as OpenAI (GPT-5.2), Anthropic (Claude Opus 4.5), and Google (Gemini 3 Flash) within a single workflow. Through design that leverages each model's characteristics for role distribution, processing stability, and output quality, you can achieve consistent results even for complex tasks.

4. GitHub Integration for Automation

Automatically execute AI applications triggered by GitHub operations such as Issues, Pull Requests, and comments. This isn't just simple automation—it enables processing that understands the context and intent of PRs and Issues. Achieve automation deeply integrated into development workflows: reviews that consider the intent behind code changes, linking to related Issues, document generation, Issue classification and handling, and more.

5. Custom Knowledge Utilization

Incorporate your own data—such as company PDFs and documents, or code, Issues, and PRs from GitHub repositories—as knowledge in your AI applications. Enable answers based on organization-specific information that general AI cannot provide. By utilizing GitHub repositories as custom knowledge, you can also perform reverse engineering to create specification documents from source code. Advanced technologies like RAG and Vector Store become accessible to anyone without specialized knowledge.

Use Cases for Giselle

Giselle Pricing Plans

The open-source version is available for free through self-hosting.

Product Hunt Launch

Starting today, December 29, 2025 (Monday) at 5:00 PM, we are launching on Product Hunt for 24 hours. Please check it out and show your support!

Product Hunt: https://www.producthunt.com/posts/giselle

Future Developments

• API provision and MCP support for created AI applications
• Secure sandbox execution for AI applications
• Expansion of external service integrations (Slack, Google Workspace, etc.)
• Expansion of supported languages and regions
• Enhanced enterprise security features
• More advanced multi-agent collaboration capabilities

We will continue to make ongoing improvements while incorporating feedback from users around the world.

About ROUTE06

ROUTE06 is a startup redefining product development through human-AI collaboration. Centered on a user experience that combines natural language dialogue with intuitive node UI, we provide AI-driven development platforms including requirements design "Acsim," AI agent builder "Giselle," and database design "Liam." Supporting the entire design, implementation, and operation process, we revolutionize development speed and quality. Leveraging our enterprise system development expertise and modern product development knowledge, we aim to create a future where all product builders—from major system integrators to startups—can freely bring their ideas to life.