New in Golden: June 2026

We shipped a lot in June.

92 updates. 31 new features, 32 improvements, 29 fixes. One month.

We won't list all of them here. Just the ones you'll experience in the product.

Two ideas drove most of it. Analytics should be as beautiful as it is correct. And it should never feel like a single-player game.

Analytics is a team sport

An analysis nobody sees doesn't create value.

So a lot of June went to shrinking the distance between "I found something" and "the right person knows."

Spaces are easier now. Create one and invite people in the same step, each with a role, with an email if you want to send one. Share an analysis out of a private space without moving it somewhere else first. Leave or remove people from any space, public ones included. That used to be blocked. It isn't anymore.

Sharing to Slack got a big improvement. Post an actual picture of the dashboard to a channel or a DM, public or private, with your name on it. And the dashboard viewer finally has a single Share button. Invite, image, Slack, export, all in one place.

Need the raw data? Download it straight from Analyze. CSV or Excel. Done.

Chat should feel like a conversation

Most people don't build charts in Golden anymore. They ask for them.

So June went to making the AI Analyst a better listener. Ask a follow-up and it changes the chart in front of you instead of starting a new one and making you find your place again. Ask for another measure and it lands on the same chart.

The insights it writes are yours to edit now. Fix the wording, fix the formatting, then share it. Try on a dozen styles without the drawer slamming shut after every pick.

It's more careful, too. It chooses chart colors that stay readable on a dark canvas. The text columns it generates, a category, a sentiment score, drop onto a shelf like any other field. And it still shows up when you're connected straight to a live database.

Your warehouse already knows what the data means

Someone already did the work. They wrote the column comments. They tagged the sensitive fields. They marked which tables are certified and which are on the way out. That knowledge sits in Snowflake and Databricks, and most tools ignore it.

Golden reads it now.

Connect Snowflake and its Semantic Views come through. Column comments, governance tags, and auto-classification like Email or Phone Number, with Certified and Deprecated badges right in the table picker. Search matches on the comments and tags, not just the names. Import a table and that meaning follows the fields into the data dictionary and into what the AI knows.

Databricks works the same way. Search Unity Catalog by name in the picker. Column tags tell Golden what a field is and how to show it, so a credit card number stays a label and never gets summed. Certified columns wear a shield. Deprecated ones get flagged, and the AI Analyst steers around them.

And when a table already has a good, human-written description, Golden uses that instead of inventing one.

Dashboards should look like you meant it

A dashboard that looks generated is a dashboard nobody trusts.

Marks can fade now. A gradient from full color to nothing, on bars, areas, scatter, treemap, combo. Cards and marks can throw a shadow, subtle up to heavy, from a single picker. Pick one Accent Color and it runs through every card and chart on the board, palettes and all. Set corner radius, border, and weight for the whole dashboard in one place.

KPI cards can wear a status badge. Green check, amber triangle, red x. You read it without reading the color.

None of this needs a designer. It needs a few clicks.

See what moved the number

Revenue's down. Now the real work starts.

Down where? Which region, which product, which segment did the damage?

The new Driver Tree chart shows you. Start with a number. Golden fans it out into the things behind it, each one sized by its share, marked against target, colored by whether it helped or hurt. Drop another field on the shelf and the next layer opens itself. Click a card to keep going.

Analyze or dashboard. Same as every other chart.

Everything else

Faster to work in

• Cmd+K from anywhere. Jump between views, upload a file, export. It knows where you are.
• More Keyboard shortcuts to move through the product faster than ever. Ctrl+1 through 9 toggle every panel.
• Your setup follows you. Templates, saved filters, field tweaks, preferences, all saved to your account. Sign out, switch machines, it's still there.
• Instant jumps between an analysis and your spaces.

Data and connections

• Per-user sign-in for Snowflake. Everyone queries as themselves, with their own credentials.
• Delete one step in Prepare and the rest rebuilds around it.
• Drag a field from Dimensions to Measures to convert it, right on the dashboard.
• More formula functions that just work: COHORT, BUSINESSDAYS, null and IS NULL, and calcs that reach across joined tables.

On the chart shelf

• Color Scale drives the gradient on filled maps.
• Outline strokes on lines and bars.
• Lasso a region on a map to select it, and cross-filter everything else.
• Click-to-filter on the specialty charts too.
• Full-dataset accuracy: density maps, box plots, histograms, Gantt, trellis, and retention now compute over every row in the database, so the statistics are right on big datasets. Dashboard trellis matches Analyze exactly.

And 29 fixes. Retention math on big datasets. A date filter that read a day early. Tooltips that now drill and format like they should. Google Sheets imports that flat-out failed. All fixed.

That's June.

92 reasons to open Golden this week. Start with one.