New in Cube: agentic analytics, one connector away from Claude. Agents don't need a better prompt. They need a harness — tools, context, checks, and a semantic layer that keeps every answer governed. cube.dev

Cube for Sheets and Excel now puts every exploration in a document-aware sidebar. Refresh all queries at once, build pivots by drag-and-drop, and catch cell conflicts before overwriting data. Full demo: piped.video/watch?v=ik1afkyi…

Cube dashboard stack containers organize related widgets without manual alignment. Use Horizontal or Vertical Stack to group elements and distribute them evenly by width or height. Full demo: piped.video/watch?v=cSuvpDY5…

Cube’s dashboard Field Switcher lets viewers change chart dimensions and measures on demand. Pair it with a Parent control to switch multiple fields across connected views. Full demo: piped.video/watch?v=fisPxGRX…

Cube record tables make row-level analysis easier: display one record as readable fields, compare multiple records side by side, and switch from dense rows to a detail-first view. Where would you use this?

New in Cube: composite charts. Combine multiple measures in one chart, then set the chart type, axis, breakdown, color, line style, and point shape for each one. See the full changelog: cube.dev/changelog/2026-09-1…

A large context window can still be overloaded. Treat agent context as a budget: page big results, label truncation, and search governed metrics on demand. Five production lessons: cube.dev/blog/building-an-ag…

Cube funnel charts make conversion drop-off easier to inspect: select a funnel view, compare each step, and spot where the largest drop-offs happen. Full demo: piped.video/watch?v=u2T8kWcf…

Agents can write SQL, but schemas do not explain how a business defines revenue, joins, or access policy. Cube argues the semantic layer is the safer interface for governed agentic analytics. cube.dev/blog/cube-connector…

Demo: ingest Cube assets into Atlan and visualize lineage from Cube metrics back through the catalog. Full demo: piped.video/watch?v=Fx02OmQI…

Cube smart data visualization selector helps pick a chart that fits the query result: inspect the data shape, switch views quickly, and get to a clearer dashboard faster. Full demo: piped.video/watch?v=ORgDS9zb…

Cube iframe events help host apps react to embedded analytics activity. Listen for message events, detect views, navigation, downloads, drilldowns, AI queries, and errors, then connect those actions back to your product workflows. What would you automate from an embed event?

Cube bins and value groups turn noisy raw values into cleaner analysis buckets: create numeric ranges, group categories, and make charts easier to scan. Full demo: piped.video/watch?v=mQexSc26…

New in Cube: Chat Artifacts keeps model branches, Workbooks, Dashboards, and saved explorations tied to each Analytics Chat—so teams can switch conversations, work in parallel, and keep context intact. cube.dev/changelog/2026-09-1…

Cube time zone controls now cover users, dashboards, embeds, and agents, so global teams can keep results aligned to the right local reporting window. Full demo: piped.video/watch?v=nzfDj5qr…

Cube Fill in Missing Rows keeps time-series charts from dropping empty dates. The demo shows one-click missing day/week/month rows, untouched source results, and zero points so chart gaps are visible. Full demo: piped.video/watch?v=We7kAnfg…

Cube dashboard chart downloads let viewers export individual charts as PNG or PDF for docs, decks, and stakeholder updates without rebuilding the visual elsewhere.

Cube results freshness indicators show how recently chart data was updated. The demo covers the signal in workbooks, dashboard builder, and published dashboards so viewers can judge recency at a glance. Full demo: piped.video/watch?v=oycWBU4b…

Five years ago, we introduced the first version of Cube Store, our query engine for building and serving pre-aggregations. It is a key component of how Cube builds smaller aggregated tables from semantic layer definitions and uses aggregate awareness to route queries to matching aggregates. The volume of queries served by Cube Store has grown exponentially since last year, as agents have become consumers of business intelligence. We've learned a lot about building, refreshing, and serving those aggregates, and how that workload differs from traditional analytical workloads in cloud data warehouses. In the blog post below, we've compiled those lessons and shared some of the architecture decisions we made to optimize serving and refreshing pre-aggregations. Igor Lukanin did a great job benchmarking Cube Store with a set of queries adapted from TPC-H to show how it behaves across query types and scale factors. Blog and benchmark: cube.dev/blog/cube-store-per…

Python analysis in Cube brings custom code into the workbook flow, so teams can run deeper analysis against Cube-powered data without exporting away from governed metrics. Full demo: piped.video/watch?v=RVn11Kue…