September's biggest changes landed in Katalon True Platform, and most of them share one idea: change should be deliberate. Test cases now change only when someone decides they should, and every version records whether a person or an AI run made it. Katalon AI works from your own requirements, documents and Jira acceptance criteria, and tells you when those sources disagree. Teams with test cases in Azure DevOps get a direct import path, dashboards now reach across projects, and TrueTest turns more real user sessions into tests.
If you read the August roundup, where test case versioning first shipped, this is the month that foundation becomes the way the product works day to day. Here's everything that shipped in September 2026.
Most of September's work landed in Katalon True Platform. Test management updates live in TestOps, execution updates live in TestCloud, and a few touch both.
The test case page has been rebuilt around one job: writing and editing test steps. Everything else, Details, Linkages, Execution and History, now sits in tabs on the right panel. Save controls live in the header so they're always within reach, and the breadcrumb shows the full folder path, so you always know where a test case sits in the repository.
The biggest behavior change is that auto-save is gone. Your edits now stay in a working draft until you save them. An "Unsaved changes" indicator shows when you have pending edits, you get a prompt before leaving the page with unsaved work, and you can discard the draft and go back to the latest version in one click. If your team is used to edits saving themselves, it's worth a quick heads-up before people run into it.
Only changes to what a test case actually does create a new version: its steps, preconditions, description and step attachments. Updating status, priority or tags doesn't, so the history stays focused on meaningful changes rather than housekeeping.
When you save, you choose how official the change is: Save as version creates a draft version, while Save & publish saves and publishes the next major version in one step.
The History tab lets you browse every version with its status, author, source and timestamp, open any past version in read-only mode, compare versions, and restore one. Restoring creates a new latest version and never overwrites history. Alongside it, a newest-first activity log records every change to the test case, showing who made it, when, where it came from and what changed, filterable by person, date range, or both.
Finally, every version is labeled as made by a person or by AI, and AI-made versions link back to the agent run that created them. As more test authoring shifts to AI, that gives reviewers a direct way to see exactly what AI changed and decide whether it should stand. Taken together, the redesign gives teams control over when a change becomes official, traceability over who changed what and when, and clear oversight of AI-authored work inside the test record.
Teams with existing test cases in Azure DevOps can now import them directly into a Katalon True Platform folder. The import brings over names, descriptions, steps, expected results, and mapped status and priority values, and since late September it also keeps each test case's Automation Status.
One thing to plan around: the import is a one-time copy. Later changes aren't synchronized in either direction, so treat it as a migration step rather than a live connection.
Dashboards now have a global, cross-project entry point, so leads who look after several projects can see them in one place instead of rebuilding the same view in each. New dashboards can start from a starter template library rather than a blank canvas. The custom chart builder also gains a Test Runs dataset, so you can chart test-run results directly.
For more on building reports that stakeholders actually read, see our guide to smart test reporting and the custom dashboards documentation.
August gave admins direct control over Feature Area, Test Type and Tag values. September extends tags beyond test cases: you can now add and manage tags at the test suite, test run and test result levels, then filter, group and report on them. Test case views can also be filtered by custom fields and tags, which makes large repositories much easier to slice by release, team or risk area.
A single test run or Test Suite Collection can now combine web, mobile-native and web-service suites, with each suite keeping its own execution configuration. If an end-to-end flow spans a web front end, a mobile app and the APIs behind them, you can run and report on it as one unit instead of stitching three separate runs together afterward.
For self-hosted schedules, a new Auto-select available agent option picks an eligible agent at trigger time, and matching rules let you narrow which agents qualify. It's a better fit than pinning a schedule to one specific machine that might be busy or offline when the run starts.
A handful of smaller updates round out the month:
For the complete list, see the TestOps release notes.
September's Katalon AI work is about context: giving AI-assisted testing your project's own knowledge, and being transparent when that knowledge doesn't line up.
Katalon AI Assistant can now ground its answers and generated tests in project-specific knowledge. You can index existing test cases, test suites, test folders and requirements, then add supporting documents such as PRDs, specifications, test plans, style guides and glossaries through Document Storage.
The practical effect is less repeated context-setting. Answers and generated tests can reflect the terminology, requirements and quality standards your team already uses, and product knowledge stays attached to the project instead of being scattered across shared drives and chat threads. It builds on the document grounding introduced in August, and pairs naturally with AI test case generation in Katalon True Platform.
Katalon AI now reads Acceptance Criteria and other mapped Jira requirement fields when it answers questions and generates tests. Acceptance criteria are usually the most precise statement of what "done" means for a story, so drawing on them directly helps generated tests reflect what the team actually agreed to build.
At the end of requirement analysis or test generation, Katalon AI now checks connected knowledge and existing tests for conflicting information. When it finds a conflict, it identifies both sources and suggests a resolution. It reports conflicts only and never edits the sources, so the decision stays with your team.
A requirement that says one thing while a test plan says another is exactly the kind of gap that tends to surface late. Flagging it while tests are being generated gives reviewers a chance to settle it first. Our guide on how to review and validate AI-generated test cases covers how to build checks like this into your review process.
Run with AI can now handle email one-time passcodes during execution. Register a MailSlurp inbox under Settings > Configurations > Secrets & Variables, then reference it naturally in a test step, for example Enter the OTP from payroll inbox. The runner retrieves a fresh code at runtime, enters it into the application, and redacts the value from test data, run history, AI logs, and API or MCP responses.
Later in the month, Run with AI also learned to fill grouped six-digit OTP fields, the split-box inputs many sign-in screens use. Together, these mean email verification no longer has to interrupt an automated journey or rely on hard-coded workarounds.
On accounts where the capability is enabled, teams can now connect Azure DevOps to Katalon AI Assistant with either Entra ID OAuth or a personal access token. The extra option makes it easier to bring Azure DevOps work items, test plans, pipelines and related context into AI-assisted testing across different enterprise identity setups.
The Katalon True Platform MCP Server adds update_test_results, requirement-ID lookup, and standardized values for create_test_run, so agents working through MCP can record results and create runs more consistently. Alongside them, fixes correct automation-status and successful-save responses, and quoted text is now preserved when agent rules are exported and imported as CSV. See the MCP server documentation for the full tool list.
TrueTest turns real user sessions into test coverage, and September made both sides of that easier: generating tests from sessions, and finding the sessions worth looking at.
You can now regenerate an existing test case directly from User Session details, including after editing a session that already has a generated test case. A single unsaved-changes dialog handles regeneration, and assertion and Delete steps are correctly disabled once a session has a generated test case, so the session and its test stay in step.
Captured browser console messages now appear alongside the Session Replay timeline, with controls for reviewing the events tied to each test step. When a journey goes wrong, you can see what the browser was reporting at that exact moment without leaving the replay.
Session filters could already be combined, but there was no way to exclude a condition. You can now add NOT to exclude an individual condition or a whole group, for example sessions from test environments. It works in User Sessions search, Journey Map generation and map-flow filtering, so your analysis focuses on the journeys real users take.
The Tracking period picker previously rejected any end date later than today. You can now choose a start date no later than today with an end date in the future, then apply that range to search available sessions. Future start dates are still blocked, and the existing start-before-end and 90-day range limits still apply.
Across Katalon True Platform, September was about deliberate change. Test cases change when someone decides they should, with a clear record of whether a person or AI made each version. Katalon AI works from your own requirements and documents, and flags disagreements instead of quietly picking a side. Azure DevOps test cases have a way in, dashboards reach across projects, and TrueTest turns more real sessions into tests. Try what's new, and tell us what you'd like to see next.