Metabot AI
A new toolbox of LLM-backed stuff to play with in Metabase. Query data in plain English. Generate and debug SQL. Automate written summaries. And get fast answers from our documentation directly in Metabase.
Let Metabot help you and your team self-serve data
Type your question and Metabot delivers answers. Forget "no SQL required" - you don't even need to click on a filter.
Shortcut your way to native queries. Write a prompt in the SQL editor and Metabot will generate it for you. Or write your own SQL and get Metabot to fix any mistakes.
Get Metabot to tell you what's going on in your monthly KPIs, or do the legwork on a writeup for you to share with your team.
Need a refresher on the difference between a left join and an inner join? Ask Metabot. It'll consult the docs and get back to you with a simple answer right where you're already working.
Click on a spike in a chart (or anywhere - but pointy stuff is usually exciting) for an automagically generated dashboard on what might be behind it.
Metabot is Metabase's built-in AI assistant that helps you work with your data using plain English. You can ask questions in natural language - just like you'd ask someone on your data team - and Metabot AI will run the query and turn results into charts. It can also help explain results, write SQL from a natural language prompt and debug SQL queries, and give you quick answers from Metabase docs. We've got more exciting AI-backed features coming soon, check out our roadmap for a peek into the future.
Metabot is available now in beta! It's included with the 14 day free trial of Pro and Enterprise plans for a limited time. If you’re already on a Pro or Enterprise plan, you can try Metabot free for one month by activating it in your Admin settings.
Yep! Metabot can be embedded as a React component with the Embedded Analytics SDK.
We’ve put a lot of time and thought into training the models and fine-tuning Metabot, but like all things AI, it’s never perfect. One of the cool things about Metabot, though, is that it's transparent with how it gets answers and results. It'll tell you which models and metrics it's referencing, and you can take a look at how it's put the question together in the query builder or SQL editor and you can tweak the query if you need to.
Yep. Metabase is a querying and visualization layer that sits on top of your database, and Metabot turns your question into a SQL query on the data in your database. For example, you can type “Show me total sales by month for the last year”, and Metabase will generate the SQL, run it, and display the results without you writing a single line of code or clicking a filter.
Yes, both. If you prefer writing SQL yourself but hit a snag, Metabot can debug errors for you. No need to copy-paste between your editor and Claude or ChatGPT or whatever the kids are using these days.
When you ask Metabot a question, like "What were total sales by month for the last year?", you can specify the chart type you'd like to see your data represented as and Metabot will gladly oblige. Or you can ask it to change the chart type on an existing visualization. You'll still need to save the visualizations to a dashboard, but that's the easiest part (and who knows what Metabot will be able to do in a month or two, the kid learns fast).
Metabot and other AI-backed features will get their start in Metabase Cloud, with self-hosted likely to follow (there's more for us to figure out with self-hosted around accuracy and privacy, we're working on it).
Unlike standalone AI analytics tools, Metabot AI is built into the full Metabase platform with careful consideration around things like your permissions structure and metadata so it only surfaces data users are authorized to see, in terms they're familiar with. It works with your existing databases and lets you embed AI-powered analytics directly into your apps.
