Genie Code: AI-Powered Development for Data Teams

Summary

Genie Code is Databricks' autonomous coding agent for technical data practitioners, used by over 90% of Databricks customers and responsible for more than 75% of all code authored on the platform — growing roughly 10x since the beginning of the year.
Unlike general-purpose coding assistants, Genie Code understands your data catalog, column usage patterns, join relationships, and team history to generate contextually accurate notebooks, dashboards, and SQL queries.
Weston Hutchins leads the Genie Code product team, which also owns Lake Flow Designer, Databricks' AI data prep tool, and sees data platform teams as among the heaviest users of the product.

Genie Code: AI-Powered Development for Data Teams

Genie Code is the autonomous agent for technical data practitioners on Databricks, used by 90% of customers to author over 75% of code inside the platform. Learn how Genie Code goes beyond code completion to autonomously create notebooks, build dashboards, generate SQL queries, and debug pipelines with enterprise context about your tables, lineage, and team assets.
Weston Hutchins walks through how Genie Code differs from general-purpose coding assistants by understanding your data catalog, column usage, join patterns, and the work your team has done before. Discover real workflows: instant data prep, dashboard creation without manual SQL, one-line edits for quick changes, and debug-on-demand troubleshooting. See how Genie Code learns from repeated patterns to surface suggestions and adapt to your team's style, plus how it integrates with Lake Flow Designer and other Databricks products.
🤝

Chapters

FAQs

What is Genie Code and how does it differ from other AI coding assistants?

Genie Code is Databricks' workspace-integrated AI agent built specifically for data scientists, machine learning engineers, data engineers, and analysts who work inside Databricks. It differs from general-purpose coding assistants by understanding your data catalog, column usage patterns, join relationships, and the work your team has already done — enabling accurate code generation without manual context-setting.

How widely is Genie Code used across Databricks customers?

Over 90% of Databricks customers use Genie Code every day, and it now authors more than 75% of all code written inside the platform. According to this video, it has grown roughly 10x since the beginning of the year, making it one of the fastest growing capabilities in the Databricks workspace.

What can Genie Code do autonomously for data teams?

Genie Code can autonomously create notebooks, build dashboards, generate SQL queries, debug pipelines, and prepare data without requiring users to write code from scratch. It also supports one-line edits for quick interactive changes and integrates with Lake Flow Designer, Databricks' AI data prep tool.

Who are the primary users of Genie Code within a data organization?

Data platform teams are among the top Genie Code users, alongside data scientists, machine learning engineers, data engineers, and analysts who build dashboards inside Databricks. The product is designed for any technical data practitioner working in the Databricks workspace who wants to increase their daily productivity.

Full transcript

[00:20] All right, we are back Summit Live. Holly, you just did an amazing demo. I did an amazing run across the conference hall if I may say so myself. Thank you. Thank you for that. That was You did a lot of work in there. But yeah, to run there I was like, "What what I'll cover for you." And we also have Weston talking about one of the
[00:36] most amazing topics that we have, you know, Genie and then Genie code as well, one of the top ones. So, why don't you say quick hello? Yeah, uh hi. My name is Weston. Thank you all for having me. Um I uh lead the product team that's responsible for Genie code and a bunch of actually
[00:52] products built on top of Genie code like Lake Flow Designer. That's like our AI data prep tool. Um and my job's really fun. I get to see how our data teams are are changing the way they work every single day using products like Genie code. All right, so help anchor us, please,
[01:07] because there's a lot of Genie stuff out there. Could you give us a lay of the land of the Genie bits and pieces Sorry, this is a loaded question. Bits and pieces of the Genie ecosystem at the moment. Well, funny funny it together? Funny story, I've actually primed Genie code to uh distinguish between all the
[01:23] Genie products when I ask it. So, that's a a big difference. That is a pro tip for you all out there. Like add that to your instruction set. But now, Genie code is really our product for technical data practitioners. So, Genie 1 and Genie is for our our business users. But
[01:38] if you live inside Databricks, you're going to be working in Genie code. This is our workspace product. We've integrated it directly into the workspace. So, data scientists, machine learning engineers, data engineers, even analysts that are building dashboards will use Genie code to make themselves more productive inside Databricks. And we were talking before that a lot of
[01:56] people like what percent of Databricks code is assisted through Genie code? Yeah, we've seen an explosion of growth. So, actually over 90% of Databricks customers use Genie code every single day. We just reached a milestone where more
[02:11] code is authored on Genie code than handwritten inside the platform. So, over 75% of code in Databricks today is authored through Genie code. It's grown about 10x since the beginning of the year. We've just seen massive, massive growth.
[02:27] It's actually really fun talking to customers about it, too, because they you know, it's one of those products when you use it, it's really hard to go back. They said, "Please never take this away from me while I'm using it." So, I think one of the key differentiators obviously is very geared towards the Databricks platform. It's very kind of it's built by data people
[02:42] for data people. Like, what do you think it has the advantage over some of the other options out there? Yeah, I mean, look, they they I use all the products, right? Like, part of my job is to understand and use these different tools. Other products out there are really good at like general purpose coding
[02:58] questions, but I I often find myself reaching for Genie and Genie code when it comes to analytical tasks. Like, things that involve data. And like some really good examples, you know, there's data there's a data scientists and data engineers on my team that create these notebooks. I
[03:13] actually will feed that into Genie code to create forecasts and and projections moving forward to look at like user churn and adoption growth. These are things where I don't even have to prime Genie code. I just ask it for that particular, you know, example and it knows how to surface those tables. So,
[03:29] you know, really Genie code is all about, you know, context and understanding your tables and your catalog and your lineage, but more so than that, just the people that you work with and the assets they create. It just knows about those out of the box. Yeah, no, I just want to say you had mentioned everyone here using it. I
[03:45] brought up I have Genie one on my iPhone. We want everybody out there, you know, it was announced today. It's incredible. I you know, this is like the bakehouse demo from the the demo before to show. But for me personally, uh I'm
[04:00] in marketing. I use it to see how's our registration going. Ask questions about Slack's and Slack's interface is so bad, but tons of applications and excited that incredible growth already this year what everyone here will be working with. Yeah. So, there's going to be a lot of
[04:16] people at home who are interested in Databricks and maybe not using it, you know, in anger at the moment. Genie code is so much more than just like a little helper. It's It's not Clippy, is it? There's so much more than just kind of like tell you about
[04:32] your data. Can you maybe talk through some things that you think for the data people out there, they're going to take one look at that and be like, "Oh, yeah, that's actually I actually need that feature." Yeah, I mean, I It was about 3 years ago I got up on stage and I introduced Databricks Assistant. We reintroduced
[04:47] everybody to Genie code, which is really the evolution. It's gone from a chatbot. It's not Clippy. It's an autonomous agent that works inside your product. So, the best part The best example I can give is go into Genie code and ask what it can do. And it will give you a list of examples, probably create a table for you. I use it all the time to create
[05:03] notebooks. I have now used it entirely to create dashboards. I don't do it by hand anymore. I don't write my SQL queries. I say, "Hey, here is my data. This is the type of information I want to get out and create a dashboard for it." It can go do all of that autonomously. Um so, really, you know, just explore.
[05:20] And honestly, try things out in it. And if, you know, you'll find out what works and what doesn't work, but it gets better every every single week. So, Yeah, I mean, so you do have the the kind of side panel and that will make changes for you. And so, especially for me, I'm quite lazy. I just match that debug button like every time it errors
[05:36] and it comes back and it does something. But I think also, when you know, like there's a one-line change, you could just highlight something and be like, "Can you make this a random number rather than a static one?" And it just does it. And for someone who makes a lot of demos, like it's a lot of dog's body work, frankly. And I don't really want
[05:52] to have to spend a lot of time being like, "Oh, can you just like mess up this data, please, so that it looks natural?" Like previously, that would have taken me like, I don't know, half a day, too long, shall we say? Whereas now, I can just say, "Can you make a table, make it look like this, and I want to do it for this, and here's the kind of demo I want to do." And it'll go, "Okay-dokey." And it'll come back,
[06:07] and I'm done, which is great. But that's kind of a I want to say trivial use. Who is the biggest user of Genie code of Databricks? Do you know? Uh so, we use data we use Genie code a lot inside Databricks. It's actually really cool to see the internal adoption. Our number one Genie code user
[06:23] is a member of our data platform team. So, all you customers that have data platform teams, our you our data platform team is our biggest user. And Holly, I I know this cuz I look at stats of our users and give who gives feedback. You're a big user. What is your favorite feature?
[06:41] Uh what is my favorite feature? Um Yeah, I I definitely think when it comes to the debugging side of it, um I think especially as I like it's a means to an end for me. Like I once upon a time I used to do like big data engineering
[06:56] pipelines and things like that. I would do like very, very specialized work, but now it's a little bit kind of more day-to-day. Um I just very easy like if I know what I want, it just kind of like helps me execute a lot faster. Um I think some of the things I have
[07:11] seen that I really like, oh, being able to tag the table that you want and having to stop playing guessing. Like that was a feature that I didn't know I needed until it arrived. Um I think also just a few like the inbuilt kind of baked in ones as well. Although I do have one one mild criticism.
[07:27] Okay, here we go. When I'm doing something and I'm trying to demo Genie, sometimes I'll like do it over and over again. But then it'll pick up that I'm repeating the same thing over and over again. It'll help me in my demo. Be like, "It looks like you query this table all the time." And I'm like, "No, it's got to look like doing it for the
[07:42] first time." It does get better the more you use it, and that's a very good example of that. Yeah. It learns. It adopts. And I got the signal that we have less than a minute left. I want to make sure people know All right, they they're ex- everyone's excited about this. How do they get started? Yeah, so uh go into the product. You'll
[07:59] open up the side panel. Uh coming soon, there will actually be a button that you can click to launch Genie code into a full page experience. Um that is going to be our new command center using Genie code. Open it up, ask it questions, see what you can do. All right, I can't wait to use it. Thank
[08:15] you so much, everyone. Thank you.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.