💬 Redesigning our AI chat

See our full changelog here: https://www.basedash.com/changelog

Edit charts directly in chat

You can now refine a chart without starting over. When you ask the assistant to tweak a chart it already made in the current conversation, it edits that chart in place instead of spinning up a brand-new replacement—so iterating on a visualization feels like a real back-and-forth rather than a growing pile of near-duplicates.

Chat also keeps each chart pinned to the exact version it referenced, so scrolling back through a conversation always shows the chart as it looked at that moment. Adding an older chart to a dashboard or copying it brings over that original version, not whatever the latest edit happened to be.

A cleaner, more capable AI chat

The chat experience got a visual refresh. Your own messages now sit in a rounded bubble aligned to the right with no avatar, while messages from teammates (and from Slack) show their name and avatar on the left—so multi-person conversations are much easier to follow at a glance.

The assistant’s thinking is tidier too. While it works, you see live tool activity; once it’s done, the steps collapse into a simple “Thought for…” summary you can expand on hover. We also unified the scroll-to-bottom button across every chat surface and fixed completed web searches so they show their sources instead of a stuck loading spinner.

Smarter analysis across your databases

Basedash’s AI analyst is now more rigorous about the kinds of questions that trip up most tools. It’s more careful with ambiguous metrics, distinguishes named rates from their supporting deltas, notices when a data source is stale versus fresh, and double-checks compound, multi-part answers before responding—so you get answers you can trust, with the caveats that matter.

We also taught it production-tested SQL gotchas for the databases you actually use—Postgres, Supabase, Redshift, Snowflake, SQL Server, MySQL, PlanetScale, BigQuery, and more—so the queries it writes run cleanly the first time, with fewer dialect-specific errors.

Set a spending limit on AI usage

Organization admins can now set an optional cap on AI overage spend right from billing settings. Once you reach your included credits plus the limit you’ve set, AI usage pauses instead of running up an unbounded bill—and admins get a heads-up email so there are no surprises.

Leave it unset for no limit, or adjust it anytime. Either way, you stay in control of exactly how much you’re willing to spend on AI in a given period.

Fixes and improvements

• Made AI-run write queries generally available—admins can enable them per data source, and every change still requires explicit approval before it runs.

• Gave organization admins full access to every dashboard, dashboard folder, chat, and automation, regardless of individual sharing settings.

• Preserved your exact SQL output column names in table charts—use SELECT aliases like AS "Customer Name" for readable headers.

• Fixed dashboard charts that could get stuck showing stale data after a missed refresh, and moved the “Updating data from…” note into the chart header spinner.

• Moved the dashboard tab auto-advance control into the tab bar and renamed it “Auto-advance through tabs.”

• Returned chart screenshots and resource links for charts referenced through the Basedash MCP server.

• Kept MCP connections signed in longer with smoother token refresh, so you re-authenticate less often.

• Stopped scheduled reports and automations from quietly modifying your saved AI context.

• Made chart images in Slack and email reports more reliable with automatic retries on transient rendering failures.