FlowMaster

FlowMaster UI
NVIDIA AI Workbench FlowMaster Project

Inspiration

FlowMaster was inspired by the need to simplify the complex process of diagram creation for developers, cloud architects, and engineers. By leveraging NVIDIA AI Workbench, we aimed to create an AI-powered tool that reduces manual effort, while making diagram generation more efficient and accessible through natural language inputs.

What it does

FlowMaster allows users to describe their requirements in natural language via a chatbot, powered by GenAI on NVIDIA AI Workbench. The system interprets these inputs and automatically generates flow charts, cloud infrastructure diagrams, or ER diagrams. It also includes multi-diagram support for easy visualization of workflows and systems.

How we built it

We built FlowMaster using a Next.js frontend to provide the chatbot interface and interactive diagram tool, combined with a FastAPI backend that integrates psycopg2, LangChain, and the Google Gemini API. NVIDIA AI Workbench was instrumental in powering the GenAI functionalities and creating a flexible, virtualized environment to seamlessly support both the frontend and backend applications.

Challenges we ran into

A key challenge was ensuring the smooth integration of NVIDIA AI Workbench with the backend, particularly in real-time generation of diagrams based on natural language inputs.
Configuring NVIDIA AI Workbench to handle the complexities of GenAI, while ensuring it could run efficiently across different environments, was a technical challenge. Ensuring smooth deployment of both the FastAPI backend and the Next.js frontend in this virtualized environment required careful resource management.

Accomplishments that we're proud of

We successfully created an AI-powered chatbot, supported by NVIDIA AI Workbench, that simplifies the diagram creation process.
Using different frameworks like FastAPI and Next.js, we were able to integrate various technologies into a smooth and interactive user interface was a major accomplishment.

What we learned

The integration of NVIDIA AI Workbench taught us the importance of optimizing workflows for AI-driven tools, especially when combining cloud services and backend systems.
We also learned how to leverage virtualized environments effectively to manage the complexities of AI model interactions and diagram generation in real-time.

What's next for FlowMaster

Moving forward, we plan to expand FlowMaster's capabilities by adding more diagram types and further improving the AI-driven insights, all powered by NVIDIA AI Workbench. We aim to deepen cloud service integration and enhance the natural language processing of the chatbot to handle even more complex user inputs.
We are currently working on integrating an RAG application that would use the documentation of Mermaid to enhance the accuracy of the results.