ML Recommendations

ML Recommendations is a direct, permission-gated optimization workspace for opportunities, savings, policies, automation, and saved views. It is available if enabled and does not appear in the primary navigation.

Availability

Open Optimization > ML Recommendations when it is enabled for your organization. If you do not see the page, use Recommendations for the primary optimization workflow or ask an administrator to confirm access.

Review opportunities

1. Use the global filters and saved views to set the scope.
2. Review opportunity rows and savings summaries.
3. Open an opportunity to inspect Resource, Account / Region, Service, Savings, Confidence, and Routing.
4. Add optional implementation or business context during review.

Create a policy

1. Open the policy create action.
2. Enter a policy name and description.
3. Choose an outcome: Auto-forward, Keep in review, Require approval, or Monitor only.
4. Configure provider, account, region, service, recommendation, confidence, savings, and environment conditions as available.
5. Choose Any environment, Production only, or Non-production only.
6. Save the policy.

The automation table includes Policy, Outcome, Conditions, Priority, Matched in Current Scope, Monthly Savings, Status, and Actions.

Operating model

Mode	Use it when
Monitor only	You want policy impact data before routing work.
Keep in review	Engineers should inspect opportunities manually.
Require approval	The change is eligible for automation but needs an owner to approve it.
Auto-forward	The recommendation is low-risk and can be routed directly to an implementation queue.

Start with monitor-only or review-required policies until the team understands match rates, false positives, and ownership routing.