Metabase

AI just went open source.  Welcome to Metabase 60 🫶

Check out all the v63 new features: buff.ly/N0f86ir

What's new in v63: Audit logs for MCP authorizations 🔍 See who is connecting AI tools to your Metabase’s MCP server. Logs include client name, user email, approval status, callback URL, and date, filterable by event type.

Docs: buff.ly/MOxgeES .

Too many categories for a pie chart, too many rows for a bar chart. That's what treemaps are for. Native viz in Metabase 63

Metabase 63: a whole lot of wishlist items, in one release ✨ Treemap charts, Two-factor authentication, more LLM providers for Metabot, and then some. 🧵↓

The full v63 release notes: buff.ly/N0f86ir.

Embed custom visualizations with the Modular Embedding SDK Build charts in @reactjs and render them natively in your Metabase and in your embeds.

@LibreChatAI Full write-up on how we got there, what broke, and what we'd tell our past selves: buff.ly/k56byk9

Our Cloud team set out to consolidate a pile of AI logins behind one managed front door. We came out the other side self-hosting @LibreChatAI, shipping two upstream OSS fixes, and running a Slack agent that answers questions about our infrastructure.

Every company sits on a mountain of data. Almost nobody outside the data team can actually use it. Marketing wants to know which campaign drove signups. Sales wants churn by region for last quarter. And every time, the answer is the same: "Let me write a query and get back to you." Someone got tired of being that bottleneck. So they built a way around it. A tool where anyone — not just the people who know SQL — can open a dashboard, ask a real question, and get a real answer in seconds. No tickets. No waiting on the data team. No learning a query language just to see a number. It's called Metabase. And it's open source. Point it at your database and it's already generating charts and dashboards, no setup marathon required. People who do know SQL get a full editor for the hard stuff. Everyone else gets Metabot, an AI that writes the query for them. Then it keeps going: build governed data models with tracked dependencies, get alerted the second a metric moves, push reports to Slack or email on autopilot, version your work like actual code, and embed the whole thing inside your own product. 48.3k stars. 264 million Docker pulls. Over 500 contributors shipping on it constantly. The core is fully open source under AGPL, self-hosted for free, forever. There's a managed cloud and enterprise tier if you'd rather not run it yourself. Either way, the tool people used to wait days for is now something anyone on the team can just open and use. Link in the comments.

making a bunch of paywall changes to @getstrappo 1. analytics revealing users having trouble getting from launch > paywall. metabase mcp is doing a ton of work here revealing how users are landing in the app.

I'll be speaking at OUTER JOIN, our first in-person Metabase conference. If you're interested in attending, shoot me a DM. I have a discount code for my data peeps. Learn more: metabase.com/events/outer-j…

@metabase Metabase MCP has been a big win, excited to try this

@frits @lennysan 💙

@lennysan The @metabase MCP in particular has been a game changer for me.

PMs: What's one specific way AI has significantly and consistently impacted your productivity? What's something you do very differently now vs. a couple of years ago that makes your work much faster/better? (Other than prototyping, and using AI to craft/give feedback on your docs/writing)

Download the skill here: buff.ly/8Haw6Sa.

We built a Claude skill that teaches you Metabase Socratically, using examples from your own data 🏫