Katalon Studio 11.5.0: Built-In AI Assertions Land Inside Your Test Scripts, No HTTP Glue Code Required

While much of the automation world's attention this week has been on Selenium's move away from CDP, Katalon quietly shipped one of its more consequential releases of the year. Katalon Studio 11.5.0 is available for all users as of August 27, 2026, and it focuses on four things test teams have been asking for: AI that lives inside the test script instead of bolted on top of it, reports that don't force testers to juggle three different output formats, better mobile artifact handling, and enterprise-grade MCP security.

AI validation without the custom HTTP plumbing

The headline feature is a set of built-in keywords that bring AI model calls directly into test case logic. Test cases can now use built-in keywords to send a prompt, with an optional attachment, to an AI model, receive the response as text, and run AI-assisted assertions on it, with the supported keywords being Ask, Assert Text, and Assert File. That matters because, until now, teams that wanted AI-assisted validation inside a test case generally had to build and maintain a custom HTTP integration to get it, and these new keywords remove that plumbing along with the maintenance burden that came with it.

In practice, this lets testers write a step that asks a model to evaluate a screenshot, a PDF export, or a chunk of returned text against a natural-language expectation, instead of hard-coding brittle string or pixel comparisons. It's a small API surface -- three keywords -- but it closes a gap plenty of teams had already patched together with their own webhook scripts.

An AI Assistant that finally remembers your project

Katalon's in-IDE assistant also gets a persistence upgrade. According to the community release announcement, the AI Assistant now remembers your context -- tell it your project conventions and working preferences once, and it can remember them across conversations and Studio sessions, so you're not repeating yourself every time you start a new chat.

Reporting, mobile, and enterprise MCP updates

Reporting gets a rework too. Katalon Studio introduces redesigned PDF reports for test suites and Test Suite Collections, with clearer layouts and details -- including error summaries, retry attempts, test listeners and hooks, multiple comments, and AI-generated failure analysis -- that were previously available only in HTML reports. That means teams can diagnose failures and share complete execution results without switching between PDF reports, HTML reports, and Log Viewer, and the generated PDFs can now ride along as email attachments.

On mobile, Katalon Studio now accepts Android App Bundle (AAB) files for mobile recording and execution against both supported local devices and remote Appium servers, so teams whose Android build pipeline produces AAB files can test that artifact directly without installing bundletool or converting each bundle to an APK first.

For enterprises wiring Katalon into their AI agent stacks, the release also expands MCP (Model Context Protocol) support with project-wide search and Microsoft Entra ID authentication. Specifically, Katalon Studio can now connect to MCP servers protected by Microsoft Entra ID authentication, letting enterprise users authorize access to internal MCP services secured with their organization's Microsoft identity without falling back to a personal access token or an MCP proxy. Setting it up just means selecting the Entra ID authentication option when adding an external MCP server; once authorization completes, the list of connected tools updates automatically.

Desktop testers get a smaller but welcome fix too: Windows desktop tests can now automatically capture a screenshot when a step fails, using the same project-level control already available for web and mobile tests, so failure evidence shows up in execution results immediately instead of requiring a local reproduction.

Finally, the release targets performance complaints from the last two versions. Katalon Runtime Engine now opens large projects with many custom keywords and BDD step definitions without the extended Antlr4 parsing and indexing delay that appeared in 11.3.0 and 11.4.0, and BDD step-definition computation also scales better, reducing delays in feature-file suggestions.

Why it matters for QA teams

Katalon Studio 11.5.0 lands the same week Selenium pushed out 4.48, and it follows Katalon's steady monthly cadence of AI-focused updates since Studio 11 launched in February with self-healing locators and AI-driven API test generation. The pattern across the automation tooling space is now unmistakable: vendors are racing to push AI assertions, self-healing, and MCP connectivity directly into the core scripting layer rather than leaving it as a bolt-on integration. If your test suites currently lean on custom HTTP calls for AI-assisted checks, or your security team has been blocking MCP connections that lack proper enterprise authentication, this release is worth a closer look.