Inspiration
Inspiration came from our experiences working with different DBSM's, from mongoDb services up until PostgreSql. We have AI tools for everything, but still no real-time database manipulation system...
What it does
It takes a simple English prompt and database credentials from the user, (even more than one) and afterwards, does everything prompt states to do, no matter the DBSM or syntax.
How we built it
We use LLM (openAi for now) to understand user's desire out of their prompt, then generate a database query, and according to the database type and query, asynchronically make a manipulation in a correct way, afterwards we serialize the data from the database table into BSON and JSON formats and send them to our frontend via endpoints. Backend - FastAPI, frontend - React, vite, typescript. LLM-openAI
Challenges we ran into
engineering the best prompt and coming up with the most universal design for data dictionary before serializing it.
Accomplishments that we're proud of
Making our tool work on different database models, mongoDb and Postgre!
What we learned
Using ATS validation for real-life problems
What's next for DatAI
Graph-based neo-4j!
Built With
- openai
- python
- react
- typescript
- vite
Log in or sign up for Devpost to join the conversation.