AI Optimize
AI Optimize turns AI activity into measurable value, and measurement into action. It arrives as two connected capabilities — Usage and Adoption — that together let you see what your AI spend is worth and act on it. Available in Teams & Enterprise plan
Usage
Usage gives you a single view of how your connected AI systems are actually being used, what they cost, and whether that spend is paying off. A top-level overview summarises your whole estate — active users, adoption rate, total prompts, and unused seats — with drill-in views by system, department, and person. It also distils prompts into use-case clusters so you can see what people are doing with AI, and a Net ROI panel converts that activity into an estimated monthly value.
Read more about Usage here.
Adoption
Adoption acts on what Usage discovers, through three kinds of outbound email program: nudges to get idle license holders started, campaigns to spread a proven use case to teammates who haven't tried it, and surveys to validate real time saved. Each is built in a guided builder with a live preview, controlled cadence, and built-in opt-out handling — and validated time-saved figures flow straight back into Usage's ROI, closing the loop.
Read more about Adoption here.
A Refreshed RexCommand Experience
We've begun a ground-up refresh of the RexCommand experience - restructuring navigation, modernising the look and feel across major features, and repairing the flows that connect them. The goal is a platform that's faster to orient in, easier to act in, and that clearly signals what matters and where to start.
The first updated pages are now live. The AI Inventory and Policy pages have been redesigned with clearer layouts, purposeful starting points, and reduced friction in everyday tasks, fewer mandatory fields, smarter defaults, and a look that better reflects the work governance teams actually do.
This is just the beginning. More core pages will be refreshed in the coming releases as we continue rolling out the new experience across RexCommand.
New Shadow AI Connector - FortiGate
The new FortiGate connector extends Shadow AI detection to FortiGate-managed network traffic, giving you another way to surface AI tool usage in environments where endpoint or Microsoft telemetry isn't available.
It draws on the same managed AI domain list as the Defender detector, so detection stays consistent across your sources, and offers a deployment option that fits the security stack you already run. Once connected, discovered AI activity feeds into your Shadow AI view alongside your other detectors, giving you a clearer, more complete picture of AI usage across your network, even where other signals can't reach.
Availlable in Teams & Enterprise plan.
Risk-Tiered Approval Templates
The new Risk-Tiered Approval Templates replace the single, one-size-fits-all AI approval gate with a framework that routes each AI system through the right approval path based on its classified risk tier. Authored in the Approval Flow admin area and applied from each system's AI Inventory → Approval tab, administrators can build multiple approval templates — each with its own gates, reviewers, and required evidence — and assign them to the Critical, High, Moderate, and Low risk tiers.
Low-risk AI moves quickly through lightweight approval paths, while high-risk and regulated systems receive deeper scrutiny with additional gates, reviewers, and evidence. Each AI system automatically inherits the template that matches its risk tier, and every gate sign-off is captured with the approver and timestamp — turning approvals into enforceable, auditable controls and focusing review effort where it reduces risk most.
A Refreshed Risk Assessment Experience
Running risk assessments across multiple AI systems and datasets now has a cleaner, modernised interface, along with two new ways to manage and distribute the work.
A new batch view groups every assessment you launch together into a single, trackable unit. See the template used, how many assessments are complete or in progress, when the batch was run, who ran it, and how it was delivered — all in one place.
You can now also assign items individually. Within a single batch, each AI system or dataset can be sent to its own recipient — such as a specific vendor or team member — to complete via a secure link, using the template you selected. One bulk run can be distributed across the right owners, rather than completed all in one place.
Read more here.
Automated Compliance
Automated Compliance takes the guesswork out of risk assessments by recommending which AI systems and datasets to assess, and against which templates. The recommendation engine evaluates your AI inventory and datasets against the compliance frameworks your organisation has adopted, then surfaces a prioritised list of assessments — ranked by risk and tagged as Priority or Recommended.
By matching the right template to the right item, excluding work that's already been done, and surfacing the highest-risk systems first, teams focus their effort where it matters most and maintain consistent, framework-aligned coverage across their AI estate.
Read more about Automated Compliance here.