Xurrent ITSM - May 2026 Product Updates

This post is a living document, updated throughout the month as new product releases roll out. Each update listed here is released to our QA environment on the date shown and typically promoted to production the following week. Along the way, customers can review changes, share feedback, and help shape what ships next. Xurrent delivers product updates on a weekly cadence to keep enhancements moving continuously.

May 21, 2026

Sera AI Studio: Configuration Workspace

Sera AI Studio gives administrators a defined pathway to a successful virtual agent rollout: write the instructions, score the templates, validate against a golden set, and adjust. It is now available in QA as the single workspace for configuring Sera AI on an account.

The "Agent Instructions" tab is where administrators personalize the virtual agent for their business: terminology, tone, behavioral rules, sensitivity handling, and the business logic the system cannot infer on its own. Instructions are written in a markdown editor with a 10,000 character limit (an error is returned on save if exceeded). Inline guidance walks administrators through a recommended structure, starting with terminology and tone, then per-service detail (keywords, hints, example questions), and finishing with explicit guardrails for what the agent should never do.

The "Golden Set" and "Template Readiness" tabs together answer the question of whether the account's underlying resources are ready for the virtual agent to use. The "Golden Set" tab focuses on knowledge: administrators define expected search results for specific prompts (run as a chosen user), then click "Run All" to score the virtual agent's search quality against the set. Each row records pass or fail, response time, and the AI response, with a pass-rate summary at the top. This is the surface administrators use to confirm the agent is retrieving the right knowledge articles, validate configuration changes before they reach end users, and catch regressions after agent instructions or knowledge articles are updated.

The "Template Readiness" tab focuses on the request templates the agent will act on. It scores how well existing templates are configured for AI-assisted service management, surfacing total templates, how many are enabled, how many have completeness gaps, and an overall readiness percentage with a count of fully configured templates. A "Completeness gaps" table lists every enabled template with per-field status (Category, Service, Reg. hints, Keywords, Description, Action type) and a missing-field count, sortable to prioritize remediation. This is the tab that tells an administrator what configuration work still stands between the current state and a virtual agent that can confidently identify and act on requests.

The "Setting" tab holds Studio-level configuration. For this release it contains a single setting, a default "Run as" user that pre-fills the user column when adding new Golden Set rows. Additional settings are expected over time.

Access is gated to Account Administrator. In this release Sera AI Studio is available on Support Domain Accounts; the Directory Account version is expected to land next week.

Settings: Licensed Users Page

A new "Licensed Users" page is available under Settings. Support Domain Accounts and accounts billed through a parent customer do not see the page. Access is limited to Financial Manager, Account Administrator, and Auditor.

The page shows a 13-month chart comparing active user consumption against contracted licensing, covering the trailing 12 months plus the current month to date. Active users appear as column bars (summed across the account, its support-domain accounts, and any accounts this customer pays for). Contracted licensing appears as a line overlay drawn from the Purchase Order line items in force for each month.

Each bar opens that month's public statement in a new browser tab, including the current month and months where licensed users is zero.

The public customer statement now also lists the Purchase Order line items in force for the period (Number, Description, Quantity) above the Usage Overview link, so the statement is useful even for months that have not yet been finalized. Non-SaaS billing statements (pay-as-you-go, used-prepaid, and so on) are unchanged.

Specialist UI: Hide Internal Notes Across the Session

The hide-internal-notes control has moved from a per-request setting to a session-wide one.

Once a specialist enables it, internal notes stay hidden on every request opened for the rest of the browser session, with a persistent visual indicator shown while the filter is active so the specialist always knows it is on. The indicator is discreet enough that its presence during a screen share does not itself reveal that internal content exists.

Previously, the same control existed but was scoped to a single request. A specialist preparing for a customer-facing review (such as a service delivery review where multiple requests are walked through on a shared screen) had to toggle each request individually, and the toggle action itself was visible to the customer. The per-request behavior has been removed as part of this change.

The toggle stays on as the specialist moves between requests and refreshes tabs, and turns itself off when the browser is closed. The setting is per-specialist and does not affect other users. Internal notes can still be written and edited normally while the toggle is on; it only changes what is shown. Turning it off brings internal notes back into view right away.

Time Entries: Activity, Rate, and Agreement Columns

Activity ID, Rate ID, and Agreement ID are now available as selectable columns in the Time Entries view.

When added to the active view, the fields are included in the view's XLSX or CSV export.

RFC Types: Directory Account Scope in Bulk Edit and Import

RFC types defined on a directory account are now fully usable from bulk edit forms and CSV imports on support-domain accounts. The request template bulk edit form now shows the "RFC type" field for directory-defined types, and CSV round-trips (export then re-import) of request templates and requests preserve the RFC type rather than clearing it.

Previously, the bulk edit forms and CSV importers only recognized RFC types defined on the support-domain account itself, so directory-defined types were either hidden from the form or silently dropped on import. The single-template editor was unaffected and continues to work as before.

The CSV importers also now raise a row-level error when an "RFC type" reference cannot be resolved, rather than silently clearing the field. Imports of unchanged exported files now round-trip cleanly.

May 14, 2026

Knowledge Articles: Review Workflow

Knowledge articles now have a built-in review signal that helps knowledge managers identify and refresh stale content without forcing an unnecessary edit.

A new account-level "Knowledge article review threshold" setting controls how long an article can go without an update before it appears in the "Needs Review" view. The threshold is expressed in months and defaults to 12.

A new "Needs Review" view has been added to the knowledge article view list. It shows articles in "Validated" or "Not Validated" status whose "updated at" timestamp is older than the threshold, sorted oldest first. Articles in "Work in Progress" or "Archived" status are excluded.

A new "Mark as reviewed" action is available on the article detail page and as a bulk action on the list view. It updates the article's "updated at" timestamp without creating a new version or changing content, so reviewed articles drop off the "Needs Review" view immediately. Each review is recorded as a distinct audit log entry capturing the actor and timestamp. Any user with edit rights on the article can mark it reviewed; no additional permission is required.

The action is also available in the REST and GraphQL APIs and produces the same result as the UI, including the audit entry.

Asset Management: Configurable Staging Queue

Asset Management now provides an account-level setting to control whether CI updates flow through a staging queue, plus support for automation rules on staged CI records.

The new "Stage updates to configuration items" setting (Configuration Management, account administrators only) controls staging queue behavior. When enabled, the staging queue is used for all CI updates that originate from an approved staged CI. When disabled, CI updates bypass the staging queue and apply directly.

Automation rules can now be defined on staged CI records, with both on-create triggers (when a new staged CI is created) and on-field-update triggers (matching the options available for standard CI updates). Within a rule, the current CI is accessible via its belongs-to relation and retains its existing values, while the staged CI is accessible as a temporary virtual copy with the proposed values applied (similar to how today and related_note are exposed in other automation rules).

The dedicated approve action on a staged CI copies the staged changes to the actual CI and can trigger any automation rules defined on the CI model. This behavior is unchanged.

Virtual Agent: Accessibility Improvements

The Virtual Agent chat in Self-Service v2 has received accessibility improvements as part of our continued commitment to high accessibility standards across the product. Screen reader users can now hold a full conversation with the Virtual Agent without assistance.

Four changes in this release:

• New bot messages are announced by screen readers automatically, via a polite live region on the message container.
• The send button is now a real button rather than a non-semantic element. It is reachable by Tab, activatable with Enter and Space, and announced as a button with a clear name.
• Each reply or option is now a real button as well. They are reachable by Tab, activatable with Enter and Space, and announced with the option's visible text as its name.
• The message input has an accessible label, so screen readers announce it descriptively even when no placeholder is shown.

Layout, spacing, and styling are unchanged for sighted users.

Time Entries: Consistent Description Limit

The Time Spent Description field now accepts up to 255 characters across every entry point where the field is shown.

Previously, the limit was inconsistent. The in-note Time Tracking widget accepted about 250 characters, but reopening the same entry in the Edit Time Entry modal silently truncated the description at 80 characters. The weekly timesheet row and the Self Service or specialist sign-off field disagreed with both of the others.

All four places where a Time Spent Description can be typed or edited now share the same 255-character limit, matching the database column width. A server-side length check enforces the same maximum for descriptions submitted through APIs or imports, so the limit cannot be silently exceeded from any client.

Translations: Action Buttons Repositioned

On the single-item translation editor, the "Save & Next", "Save", "Cancel", and "Skip" buttons now sit directly beneath the translation text field, right-aligned, rather than at the bottom of the page.

Previously, the action buttons were anchored to the page bottom regardless of content size. On large monitors, this created a long mouse travel from the translation content to the save action for every entry, which slowed down high-volume translation work. The bottom-anchored layout optimized for content-heavy pages but penalized the translation editor, where the content is small and the buttons ended up disconnected from the work area.

Button ordering, behavior, and keyboard tab flow are unchanged. The change applies to every entry point that opens the single-item translation editor. Bulk translation flows and import flows are unaffected.

May 7, 2026

Specialist UI: Inline Rating Replaces Modal on Completed Requests

The rating prompt for completed requests now appears inline in specialist, matching the experience in self-service.

After clicking thumbs up or thumbs down, a short prompt appears above the note editor (encouraging for thumbs up, asking what could be improved for thumbs down).

While rating, the editor is focused on public feedback: the Internal Note tab, email tabs, and time-spent row are hidden. Rich text formatting, inline images, and file attachments remain available.

To save a rating with feedback, users type a note and click "Post". To skip the comment, they click "Skip" (thumbs up) or "Cancel" (thumbs down), which clears the editor without saving. "Reopen" (thumbs down) still requires a note, since reopening a completed request is a meaningful change.

Asset Management: CMDB Health Naming and Layout Refresh

The CMDB Health dashboard has been refreshed with clearer, more consistent naming as part of preparing the section for general availability.

The four KPI tiles and their section headers have been renamed: "Potential Duplicates" is now "Duplicate candidates", "Stale Records" is now "Check-in freshness", "Field Completeness" is now "Field population", and "Orphan CIs" is now "Relationship coverage". "Supporting Views" is now "Lifecycle & financial".

The "Configuration Items by status" tile (formerly "CI By Status") has moved from Check-in freshness into the new Lifecycle & Financial section, where it joins "Configuration Items by warranty status", "Configuration Item age distribution", and "Total Configuration Item value" as descriptions of the estate rather than its health.

Tile names now use sentence case with the full "Configuration Item(s)" instead of "CI(s)", and lowercase "by", as in "Field population by Configuration Item type".

REST API: People and Relationship Endpoints Accept Node IDs

The REST API now accepts node IDs (the base-64 strings returned by the GraphQL API) on the People endpoint and on relationship endpoints. If you pipe GraphQL results into REST calls, you no longer need to convert IDs first. This matches how every other record type already works.

Two cases were previously broken and have been fixed:

• People endpoint. A shortcut for "give me the current user" treated any non-numeric value as a request for the caller. Because node IDs are base-64 strings starting with letters, node-ID lookups were silently returning the caller instead of the intended person. The shortcut is now limited to three explicit values: empty, 0, and "me" (which is now a supported alias). Any other ID flows through to a normal lookup.
• Relationship endpoints (for example, attaching a calendar to a holiday, or linking a user to a configuration item). These take two IDs in the URL: one for the parent record and one for the related record. The parent ID was already handled correctly, but the related-record ID was being read as a plain integer, which silently turned every node ID into zero. Both IDs now accept node IDs and numeric IDs interchangeably.