Diagra

• 

  Initial Screen

• 

  A flow chart generated for the prompt "Binary Search"

• 

  A diagram explaining the black hole formation process

Inspiration

While working on assignments in our freshman year, there were numerous times where we had to create a flowchart from scratch. Creating flowcharts can be time-consuming as we have to compress everything into a single diagram. We surfed through the internet to find such a tool but couldn't find any functional ones. For this reason, we thought that it would be really handy to have an AI tool that could quickly generate diagrams.

What it does

Diagra is a very useful tool to generate diagrams and flow charts from natural language prompts. With the help of Diagra a flow-chart diagram can be generated almost instantly. And the process is really simple, just type your prompt and hit the send button.

How we built it

Diagra has been created with the help of CrewAI and LangChain.

In Diagra, under the hood, we have two agents which are created using CrewAI and LangChain The first agent receives the prompt from the user, understands it and then explains it to the second agent who then generates a GraphViz diagram back to the user

With the help of Django and Django REST framework we created an API for passing the diagram to the interactive UI which was built with the help of React and Tailwind CSS

Challenges we ran into

Initially we planned on keeping one agent. However, it was not generating the output in the expected format,thus failing to provide accurate diagrams. We fixed that using more sophisticated instructions and created a secondary agent which picks up the prompt and then explains it to the initial agent. In this way, the accuracy of the diagrams increased.

Accomplishments that we're proud of

Completing the project and intertwining everything within the provided time has been really challenging for us. Although we had a slow start,we quickly tracked our way back. What we are hoping from this project is that it will serve as a powerful tool for individuals ranging from students to someone who is just trying to grasp new concepts/possibly summarizing key points and instructions

What we learned

We learnt about AI agents and their usage. We are looking forward to find more use cases to make applications utilizing it to make the day-to-day life of people easier

What's next for Diagra

Last but not the least, we don't stop here. We dream of it to create Animated Diagrams, Pie Charts, from natural language input and much more.

NB. The demo is in English language

Built With

• crewai
• django
• django-rest-framework
• langchain
• python
• react
• tailwind