AI powered SQL editor

My colleagues have been recently working on a brilliant new upgrade to the Supabase Studio Dashboard; the SQL editor is now powered by Supabase AI.

This is has now become part of my daily workflow. I no longer need to switch back and forth between ChatGPT and Supabase, I can now, within the editor, just pop in a question, and Supabase AI will amend my query.

I think my favorite part is that after a while, I realized that my schema was being understood by Supabase AI. It was starting to reference my other tables and columns in my schema that were not present in the query I was working on (🤯).

Hey everyone! 👋

Today, we're releasing a big improvement to our SQL Editor: an AI assistant, embedded directly into the editor.

What can you do with Supabase AI?

- 🚀 Generate tables, reports, and more using natural language - 🔧 Tweak and massage your query using natural language until you’re satisfied - 🐞 Debug errors using AI - instead of cryptic messages, get direct solutions - 🔤 Title your SQL snippets automatically based on the contents of the query

✨ How does it work?

When you spin up the SQL editor, you are now presented with a brand new input where you can ask Supabase AI to build any SQL query you like - using only natural language. For example “build a table to store employees” will produce a CREATE statement auto populated with common fields that you’d likely want to track for each employee.

Of course you don’t have to stop there - Supabase AI is aware of the SQL snippet in the editor and can modify it for you. For example, you could ask Supabase AI to split a “department” text column into its own table, all while adding proper foreign key references and following best practices.

This is all presented via a new diff editor that shows which changes Supabase AI is making to your SQL snippet. This will feel exactly like a Git diff view in an editor like VS Code. Nothing is applied until you accept or discard the change. From there you’re free to keep asking Supabase AI to make changes until you're satisfied.

🎯 Motivation

We've wondered for a long time how to make it easier to teach developers how to use SQL, while at the same time speeding up development for those who do. It's fortunate we didn't solve this problem too quickly, as it turns out that AI does a much better job than we could do ourselves.

We first introduced Supabase AI into our documentation website (aka the revival of Clippy 📎 😄) to help developers answer their questions directly, ChatGPT style, instead of searching for it through the docs. If you’ve had a chance to play around with large language models (LLMs) like OpenAI’s GPT 3.5/GPT 4, you’ll know they are also incredible at generating and interpreting code. The SQL editor was the next natural fit for this technology.

With Supabase AI, you won't even need the whole weekend to scale to millions. Head over to the SQL Editor and give it a try!

📈 What’s next?

In the coming months, we're looking to sprinkle Supabase AI through more parts of the Studio. With Postgres under the hood, there's so much we can do with SQL and a little bit of AI to help you move fast. Keep an eye out for the Supabase AI icon, you never know where it will pop up next.

🌐 Remember - it's all open source!

View all the code right now on GitHub.

—-

Thanks everyone! We would absolutely love to hear your feedback in the comments.

Greg, Supabase team

Impressive launch! Simplifying SQL for non-experts is a game changer indeed. How does the Supabase SQL editor handle complex queries and maintain accuracy? I'm curious about its capabilities!

We are also launching a AI powered analytics copilot on 17th, would love your feedback.

Super excited that this is now available for everyone to use and leverage on - as a developer who's not the most familiar with SQL this is a huge time saver and also a potentially great learning tool! No longer will I need to search for "how to do xxx" on Google and try to find the best answer, all of this can be done without leaving the SQL editor ✨✨