We can now save customized filters for the agent list. This allows users to save and retrieve custom filters. This feature allows you to

• save a new filter
• update an existing filter
• rename a saved filter
• delete a saved filter.

We've added new filtering options to the Agents list. You can now filter your agents by Owner, Data Source, whether the agent is private or public, and whether the agent has Sub-Agents or Triggers configured. These filters make it easy to quickly narrow down your agent list and find exactly what you need.

Prior to this release, any images or other assets that were apart of the Agent's convo were not publicly accessible, and unaccessible to static websites generated by the Agent. Now, these images and other assets can be apart of the static website.

As an example, this is helpful when creating a customer report via the Static Website tool, and being able to include headshots of the client services team managing the account, as well as the account's logo.

We've cleaned up the AI model selection in Agent Builder to make it easier to choose the right model for your agents — and to keep you on the latest and best-performing options.

What changed:

• Removed older models that were either soon to be deprecated by providers or there's an updated option in the model series available.
• Automatically migrated all agents using older models to newer, more capable alternatives

The models you're using now:

• Agents driven by Claude Opus 4.5 → reassigned to Claude Opus 4.6
• Agents driven by other removed Anthropic model → reassigned to Claude Sonnet 4.6 (also released today as an option, and now the new default for newly created Agents!)
• Agents driven by older OpenAI models → reassigned to GPT-5.2 - Thinking

You'll still have access to several other powerful models across Anthropic, OpenAI, and Google in the Agent Builder, but they'll be the most recent, reliable and intelligent models to choose from. See more details here on what models are now available in NinjaCat: https://docs.ninjacat.io/docs/model-selection

In the Reports AI Insights and Executive Summary widget, if you previously had an older AI model selected for your dashboards, reports, or widgets, the system now automatically uses GPT-5.2 Instant — no manual changes needed. Everything continues to work seamlessly, and you'll get better results from the newer model. Nothing to do on your end; we just wanted you to know your AI features got a quiet upgrade.

We've updated the AI model options for AI Insights and AI Executive Summary to reflect OpenAI's latest and greatest. You'll now see GPT-5.2 series models (Thinking, Instant, Mini, and Nano) in the model selector. Older models have been retired from the picker — if you had one selected, you'll be automatically upgraded to GPT-5.2 Instant for the best balance of speed and quality. No action needed on your end.

Related Jira Ticket: TP-652

We've made significant improvements to our FTP and S3 data connectors to improve sync reliability, visibility, and data protection.

What's Improved

More Predictable Sync Behavior

• Manual syncs now behave consistently and follow your configured file selection settings.
• Clear visibility into whether new data was ingested or if no updated files were found.

Clearer Sync History

• Sync records are now grouped and easier to interpret.
• Each sync shows the date range of the data that was ingested.
• Improved status messaging ("in progress" vs "completed").

Stronger Data Validation

• If required columns (such as account or date fields) are missing from a file, the sync fails safely with a clear error message.
• Prevents unintended data overwrites or deletions.

Improved Handling of Non-Numeric Values

• Metric columns containing values like empty strings or "NA" will no longer cause ingestion failures.
• These values are now safely treated as null.

More Reliable Data Flow

• Non–account-matched FTP/S3 data now loads correctly into datasets.
• Sync processing has been migrated to a more robust ingestion system designed for improved scalability and monitoring.

What This Means for You

• More reliable FTP/S3 ingestion
• Better visibility into what was processed
• Safer data updates

Whats New? We've added the ability for our system to automatically scale up computing power when needed. If an export starts to time out, the system will automatically retry it using a larger warehouse, giving it more processing capacity to complete successfully.

How does this help you? • Fewer failed exports — Large or complex exports are far less likely to time out • No action needed from you — The system handles retries and scaling automatically • Faster resolution — Exports self-heal in real time instead of requiring a support request

What's New: Two new audio metrics have been added to the Google Campaign Manager 360 connector:

• Audio Completions — The total number of rich media audio completions • Audio Completion Rate — The rate at which audio plays are completed, calculated as Completions ÷ (Plays + Replays).

What's New: We now have a new dimension of "Email Subject Line" now available in our Site Impact API Integration.