Show HN: Delphi – A Natural-Language-to-SQL AI for the Modern Data Stack

Hi HN!

Before my current job, I worked in data analytics and spent 3 years leading data at an ed-tech nonprofit.

A big problem we had was "self-service" data. Everyone wanted to use data and make data-driven decisions, but it was often too complicated. So our data team spent a lot of time answering ad-hoc questions and requests to "do a quick pull."

Tons of data teams face this problem. It's the source of a bunch of memes and jokes in the community.

So I'm super excited to share Delphi.

Delphi takes plain English questions like "how much money did we make from gift cards in California last week?" and answers them using the data in your company's data warehouse.

We're using OpenAI's Codex, but unlike some of the text-to-SQL products you might have seen, Delphi integrates deeply with the Modern Data Stack tools teams are already using.

Instead of just generating arbitrary SQL, this lets Delphi scale to real-life data sets. Plus, users can use Delphi as a jumping-off point to explore data in their favorite BI tool.

Our prototype, a Slack bot, is ready for you to try today! And if you're using dbt's Semantic Layer, you can run queries against your own data. Check out the website for more, and reach out if this is a space that interests you!

Thanks everyone <3 Michael