Inspiration
QA isn't optional; it's infrastructure protection. Software failures cost between $2.4 and $2.6 trillion every year, and a single mistake can crash millions of systems in minutes. We wanted to build a "one-click" solution that creates safe, isolated spaces where AI does the heavy lifting for you. By using purely synthetic data, we make the testing process faster, easier, and much safer for your projects.
What it does
Instant Sandboxes: Provisions isolated Docker containers for demos or QA on-demand.
AI-Powered Data: Uses LLMs to generate fake actions and data in real-time, so no real data is ever used.
Visual Tracking: Features a live node graph that shows the "lifecycle" of your testing environment—from launch to destruction.
Capture & Replay: Users can record a specific scenario and "replay" it later to show off features or catch bugs.
How we built it
Backend: Docker for containerization and LLMs for synthesizing fake data in real-time.
Frontend: React with a bento-box layout for the dashboard.
Visuals: ReactFlow to create the interactive, glowing node graph.
Logs: A dark-themed monospace terminal to stream real-time agent actions.
Challenges we ran into
Managing the lifecycle of ephemeral containers while streaming real-time logs from an AI agent was tough. We had to ensure the Visual Lifecycle Graph stayed synced with the actual state of the Docker containers to provide a seamless "Observability" experience.
Accomplishments that we're proud of
We are most proud of building a Visual Lifecycle Graph that tracks AI agents in real-time , achieving zero data risk through 100% synthetic LLM generation , and creating a professional dark-mode interface that makes complex backend orchestration look simple.
What we learned
We learned how to prompt LLMs to act as consistent "Agents" within a sandbox. We also discovered the power of synthetic data for high-fidelity testing without the legal headache of PII (Personally Identifiable Information).
What's next for Demoforge
We want to add more Presets for different industries (like healthcare or retail) and improve the Capture & Replay engine to allow for even more complex QA scenarios.
Built With
- docker
- figma
- ibm-watson
- node.js
- ollama
- openai
- react
- reactflow
- typescript

Log in or sign up for Devpost to join the conversation.