Agent Conversation now supports richer @ mentions for precise entity selection. Previously, typing @ in a chat surfaced only the columns and fields available in the agent's assigned datasets. With this release, @ is expanded to also include accounts and datasets — opening a single searchable dropdown across all three entity types. Pick one and it appears in your message as a styled pill with a hover preview.

The goal was to make exact account names directly referenceable by the user. When you select an account from the @ dropdown, the agent becomes aware of both the exact account name and its underlying account ID — making it a far more reliable match when looking up data, compared to fuzzy-matching on a name alone.

You can add multiple accounts or entities to a single message, and the dropdown respects your accessible accounts list — accounts you don't have access to will not appear as options. Pills also render in conversation history and survive copy/paste.

Chart widgets now include a "Show Grid Lines" toggle in the Metric Axis tab of the Axes Settings panel. Turn it off to hide the horizontal grid lines behind chart data for a cleaner look; leave it on (the default) to keep the existing appearance. The toggle is available on all chart types except pie charts, and only appears when the metric axis is visible. Existing charts are unaffected.

Agencies using BI Connect can now refresh their Snowflake share on demand. A new "Refresh Shared Data" card on the BI Connect screen lets users rebuild the Snowflake views for their organization in one click — useful when shared data looks stale, recently added datasets are missing, or view names appear out of date. A confirmation prompt warns that active BI tool connections may see a brief interruption while the views are recreated. Underlying source data is never modified — only the Snowflake view definitions are rebuilt.

Renaming a dataset in NinjaCat now also correctly updates the corresponding view name in the Snowflake reader account. Previously, renamed datasets kept their original view names in Snowflake, which could cause confusion in connected BI tools.

You can now delete Data Cloud Connectors and their Datasets directly from the Data Cloud UI — no more filing a support request for cleanup.

What's new

Delete a Connector from either the Connector list or the Connector detail screen. Delete individual Datasets from the Connector detail screen.

Each deletion opens a confirmation modal with a "Delete associated datasets" toggle:

• Toggle off (default): The Connector is deleted, and its Datasets are preserved — they are detached from the Connector and remain available in data management as Static Datasets.
• Toggle on: The Connector and all of its associated Datasets are permanently deleted.

A typed confirmation phrase is required before any deletion proceeds, helping prevent accidental removals.

Static Datasets

When a Connector is deleted but its Datasets are retained (toggle off), those Datasets become Static Datasets — Datasets no longer attached to a Connector. Static Datasets:

• Continue to appear in data management and remain available for reporting.
• Retain all historical data from the time of detachment.
• No longer receive new data ingestion.
• Hide Connector-only actions that no longer apply.
• Surface the reason ingestion has stopped, so it's clear why the Dataset is no longer updating.

Why it matters

Previously, removing a Connector or Dataset required a support request. This change puts cleanup entirely in your hands — with built-in safeguards to ensure you don't lose data accidentally.

Agent Conversation now supports richer @ mentions for precise entity selection. Previously, typing @ in a chat surfaced only the columns and fields available in the agent's assigned datasets. With this release, @ is expanded to also include accounts and datasets — opening a single searchable dropdown across all three entity types. Pick one and it appears in your message as a styled pill with a hover preview.

The goal was to make exact account names directly referenceable by the user. When you select an account from the @ dropdown, the agent becomes aware of both the exact account name and its underlying account ID — making it a far more reliable match when looking up data, compared to fuzzy-matching on a name alone.

You can add multiple accounts or entities to a single message, and the dropdown respects your accessible accounts list — accounts you don't have access to will not appear as options. Pills also render in conversation history and survive copy/paste.

Chart widgets now include a "Show Grid Lines" toggle in the Metric Axis tab of the Axes Settings panel. Turn it off to hide the horizontal grid lines behind chart data for a cleaner look; leave it on (the default) to keep the existing appearance. The toggle is available on all chart types except pie charts, and only appears when the metric axis is visible. Existing charts are unaffected.

Agencies using BI Connect can now refresh their Snowflake share on demand. A new "Refresh Shared Data" card on the BI Connect screen lets users rebuild the Snowflake views for their organization in one click — useful when shared data looks stale, recently added datasets are missing, or view names appear out of date. A confirmation prompt warns that active BI tool connections may see a brief interruption while the views are recreated. Underlying source data is never modified — only the Snowflake view definitions are rebuilt.

Renaming a dataset in NinjaCat now also correctly updates the corresponding view name in the Snowflake reader account. Previously, renamed datasets kept their original view names in Snowflake, which could cause confusion in connected BI tools.

A new Google Ads dataset template, "Conversion and Attribution by Ad Group," is now available in the dataset builder. You are now able to report on Ad Group level data with the introduction of the Ad Group Name dimension.

You can now hide the grid lines in the template editor. A new "Grid On/Off" toggle sits in the bottom toolbar of the template editor, next to the snapping toggle — click it to hide the grid for a cleaner editing canvas, and click again to bring it back. The setting only affects the editor view. Exported reports, generated reports, and dashboards are unaffected, and grid lines remain on by default.

When you create a new Agent in NinjaCat, its Organization Access now defaults to Private instead of Can Edit. New agents stay visible only to you until you choose to share them — useful while you're still experimenting or iterating. When the agent is ready, you can change Organization Access to Can Edit (or another level) from the agent's permissions settings.

This change applies only to newly created agents. Existing agents keep whatever Organization Access they already have — nothing is changed retroactively.