Lorel.aiDocs
English

Updates

Last updated: 2026-10-30

  • You can now subscribe through AWS Marketplace

    You can now subscribe to Lorel.ai through AWS Marketplace.

    • Subscribe to the Standard plan on a monthly contract from the Lorel.ai product page on AWS Marketplace
    • Your usage charges are included in the bill of the subscription AWS account (the AWS account you used to subscribe)
    • The first time you subscribe, an initial free credit is granted

    For how to register after you subscribe, billing and payments, and how this differs from a direct contract, see Subscribing Through AWS Marketplace.

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • GitLab™ can now be configured

    We’ve expanded the external services you can connect to.

    • You can receive events from your GitLab projects by webhook
    • You issue a Webhook URL in the UI, then paste it into your GitLab webhook settings
    • You can choose which event types to receive (Push, Tag push, Merge request, Pipeline, Deployment)

    GITLAB is a trademark of GitLab Inc. in the United States and other countries and regions.

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • You can now receive notifications by email

    You can now receive notifications by email. Emails are sent to the email address you use to sign in to Lorel.ai.

    • Some notifications have email turned on by default. Only notifications that need immediate action or have a large impact if missed are included
    • You can check and change which notifications are on in Notifications (Notification Settings) under Account Settings

  • You can now choose which notifications you receive

    You can now choose which notifications you receive from Notifications (Notification Settings) in Account Settings.

    • Settings can be set per scope: Global, Organization, and Project. Lower scopes can also inherit from higher ones; for example, a change in Global can apply to your organizations and projects at once
    • You can choose whether to receive notifications for a whole category at once or for each event type
    • You can turn in-app notifications (the bell icon in the top right) and email on or off separately

  • You can now collect Amazon CloudWatch metrics

    We’ve expanded the external services you can connect to. You can set up Amazon CloudWatch metric collection from Sources under Integration, and the collected metrics appear on the Metrics page under Observability.

    • Setup runs as a four-step wizard: Access, Verify, Confirm permissions, and Collection targets
    • Metrics that actually exist in the AWS environment connected via Access are discovered for you, and you pick what to collect from them
    • Before creation, you review and acknowledge the full scope of the permissions being granted (CloudWatch read access)
    • The retention period for collected data can be set per collection target (leave it blank to use your plan’s default). The range you can set depends on your plan

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • A new Metrics page for browsing and comparing metrics

    Observability has been added to the sidebar. Its Metrics page lists the metrics collected from your connected clouds and lets you explore them over time.

    • Metrics are shown as cards with sparklines. You can search by name or tag and mark favorites
    • Selecting a card opens a detailed time series. Statistics such as minimum, maximum, average, and standard deviation are shown, and the data points for the displayed period can be exported as CSV
    • Metrics can be overlaid on a single time axis for comparison — up to 5 plotted at once, with up to 10 held as candidates. For series with different units or magnitudes, switching the vertical axis to “Change” makes them comparable on the same scale
    • Runbook and AI agent executions can be overlaid on the chart, so you can line up when a value moved with what was running
    • Each metric’s collection state is shown in both the list and the detail view: collecting (up to date or delayed), awaiting first data, collection stopped, no collection target, or connection error. Sorting by collection state brings the ones needing attention to the top
    • How far back you can look is governed by the data retention period of your plan

  • Metric thresholds can now trigger detection

    Detection rules now support metrics. An incident can be raised when a collected metric crosses a threshold.

    • Choose the metric to watch and how to narrow it down (dimensions), then set how the evaluation window is aggregated (average, maximum, minimum, median, p90 / p95 / p99), the comparison, and the threshold
    • When the target covers several series, you can decide by their average or by how many series crossed the threshold
    • You can choose how missing data is treated (as normal or as breaching) — useful when you want to detect that collection itself has stopped
    • Identifier labels let you evaluate targets separately, and the page shows how many targets that produces
    • Before saving, press “Preview” to re-evaluate the most recent period with your current settings and see the result
View 10 more

  • A single rule can now separate its detections by target

    Detection rules now have a “Target Distinction” setting. By choosing identifier labels (an instance ID, an actor, and so on), one rule can keep its detections separate per target. A single “CPU utilization above 80%” rule can now tell you which instance crossed the line.

    • When you choose identifier labels, the page previews how many targets your recent events split into
    • Rules that detect on a number of occurrences within a period now count occurrences per target
    • Compound rules also split their judgments by the identifier labels you choose
    • Incident policies have a “How incidents are split” setting: one incident per target, or one incident per trigger rule as before
    • Trigger details now include a “Target and Evidence” section, so you can see which target fired
    • The default is “Merge into one” (one incident per trigger rule). Unless you set identifier labels, behavior is unchanged

  • You can now control when AI agents run and what they can see

    You now have finer control over when a custom AI agent runs and how much it is allowed to see.

    • When an agent runs is now set on the incident policy. Choose from incident created, resolved, or reopened, and assign a different agent to each of these events
    • You can restrict which regions an agent may look at. Set a default region, then allow only the additional regions you want — useful when you need to limit data residency
    • You can set how personal information is handled when it is found, separately on the way in (before it reaches the agent) and on the way out (before results are stored or displayed). Either mask just the matching parts, or stop the processing entirely

  • You can now set a user icon

    You can now set your own user icon from Account Settings.

    • PNG, JPEG, and WebP images up to 5 MB are supported
    • Drag the frame to reposition and drag its corners to resize before saving
    • You can delete your icon at any time, which restores the default display

  • MongoDB Atlas can now be configured

    We’ve expanded the external services you can connect to.

    • You can receive MongoDB Atlas alerts by webhook
    • You issue a Webhook URL and Webhook Secret in the UI, then paste them into your MongoDB Atlas webhook settings
    • You can choose which event types to receive

  • Custom AI agents are now available

    In addition to the built-in root-cause analysis agent, you can now create custom AI agents — give each one its own name, analysis instructions (prompt), and scope of reference data.

    • Agents can use information from AWS environments connected via Access, with read-only access, as part of their analysis
    • If you specify when a Runbook should be run, the agent can propose one after analysis, and the Runbook runs once you approve it
    • You can set a rate limit on how many times an agent runs
    • You can run a “test execution” before going live, with no side effects (test execution limits vary by plan)

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • We've added a Cloud Action step to Runbook

    Runbook now has a new step type, “Cloud Action”. Using an AWS environment connected via Access, you can run AWS management API operations as a step — for example, restarting an EC2 instance, redeploying an ECS service, or restarting an RDS instance (AWS support comes first, with other clouds planned).

    • You can search for and select the service and operation right in the UI
    • What you can run is limited to operations that fall within both the operations selected on Access and the permissions of the connected IAM role (broader IAM permissions do not extend beyond the operations you select). Operations that write or delete resources can run if you select them on Access, so review both the selected operations and the IAM permissions carefully

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • Runbook is now available

    You can now combine 13 step types (command execution, conditional branching, approval steps, loops, and more) to build complex operational procedures without writing code. You can even run arbitrary commands on Amazon EC2 instances in AWS environments connected via Access.

    • You can now cancel a running Runbook partway through. If something isn’t going as expected, you can stop the running step, and you can receive notifications about the results. Results of steps that already completed (including side effects such as command execution or API calls) are not rolled back
    • We’ve expanded the simple mode of the condition builder, letting you compose more flexible branching right in the UI
    • Command / Script steps can now extract variables from their output, making exit codes and standard output available to later steps
    • Step output files can now be downloaded securely
    • AI agent definitions have been expanded to support attachments
    • Incident policies can now be set so that Runbooks and AI agents are mutually exclusive

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • Access is now available

    In the new “Access” section under Integration, you can now securely issue and manage temporary AWS access (AssumeRole). This is where you manage, in one place, the connections used by Runbook and other features.

    • You can now manage an allowlist of connection destinations from the screen
    • We’ve made UI/UX improvements and bug fixes

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • Datadog and PagerDuty can now be configured

    We’ve expanded the monitoring tools you can connect to.

    • Datadog can now be configured in the UI
    • PagerDuty can now be configured in the UI

    Other company, product, and service names mentioned on this page are generally trademarks or registered trademarks of their respective owners. For details, please see Trademarks.

  • Notifications are now available

    You can now receive notifications via the bell icon in the top right for the following events:

    • Incident created, assignee changed, severity changed, reopened, and mentions of you
    • Task created, assignee changed, severity changed, reopened, and mentions of you
    • AI agent execution failures or skips

    Click the bell icon to view your notification list, mark items read/unread, delete individually, or mark all as read.