🎙️[Live now] Ask Katalon Anything (AKA): Rethinking Test Recording & Designing Agentic Workflows

[arvind.choudhary]

@xuan.tran
As Katalon already detects the duplicate test objects, so my another question in sync to thatis below ….
Are there plans for an agentic workflow that can scan the entire Object Repository, detect duplicated test objects, unify similar locators, and refactor test references across the project automatically?

[arvind.choudhary]

@xuan.tran
My another question is around Autonomous Test Healing

Is Katalon planning to expand AI-driven self-healing into a full autonomous workflow — where AI not only proposes a fix but can apply, validate, and version-control those fixes automatically?

[arvind.choudhary]

@xuan.tran
As AI begins creating more tests automatically, what governance features will ensure those tests are transparent, traceable, and compliant with regulatory standards—for example, in industries like finance, healthcare, or insurance?

[arvind.choudhary]

@xuan.tran
Do you envision agentic workflows supporting adaptive, AI-driven runtime decisions—for instance, dynamically choosing different test paths in response to application behavior?

[arvind.choudhary]

@xuan.tran
My question is around Automated tests which often break whenever the app’s screens or buttons change. Frequent UI changes create significant maintenance overhead in automation. How do agentic AI workflows aim to solve or reduce this burden?

[arvind.choudhary]

@xuan.tran
In the future, how does Katalon plan to use AI in testing?
Do you think AI will simply help humans test software, or will it eventually be smart enough to design and run tests completely on its own?

[nghi.phan]

Hi Arvind! We really appreciate your enthusiasm and engagement on AKA You’ve now submitted 20 questions, so to keep things fair and leave room for others to participate, we won’t be accepting additional questions from you in this thread. Thank you for your understanding and for contributing! Please kindly wait for our PMs to reply. Thank you for your enthusiasm as always.

[arvind.choudhary]

Sure Bella,

Noted.
Although I am not stopping others to post questions, okay I will not post any questions further

Regards,
Arvind Kumar C

[sneha.sajeev]

@xuan.tran Many teams rely heavily on data-driven testing for scenarios like login, payments, and negative flows.
How can AI or agentic workflows improve data-driven testing in Katalon without increasing script complexity?

[xuan.tran]

@sneha.sajeev I think DDT is still a good approach. Now with AI, you can generate a test data file in seconds by just typing in what you need into StudioAssist chat. You can also attach the test case for more context and less explanation. No longer have to manually prepare test scenarios and test data yourself, it’s much easier now with AI

[sneha.sajeev]

@duy.lam UI testing often goes beyond functionality (spacing, punctuation, labels, visual consistency).
Are there any plans for AI-assisted UI validation in Katalon to help testers catch UI perfection issues automatically?

[snehass]

@Shin For QAs who are not strong in programming and prefer low-code or minimal scripting,
how does AI-powered test recording reduce maintenance effort while still keeping tests reliable for long-term projects?

[xuan.tran]

@arvind.choudhary

“Do we have any plan where Katalon support orchestrating multiple execution agents so that they can cooperate in running different layers of validation—such as Web UI interactions, API calls, and Database checks—within one intelligent end-to-end workflow where each agent can share data, synchronize steps, and run as a unified automated process?”

@arvind.choudhary I cannot share the roadmap here, but this is definitely something you will see soon So stay with us and wait for the good new stuff coming

[xuan.tran]

@arvind.choudhary

When will search history be available in Studio Assist? Having a record of our past queries would help us refer back to previous conversations and continue interacting with the AI in context, without needing to start a new session every time.

It makes me agitated if I need to research or debug deep down further in any topic which I left open couple of days earlier. Now I need to again begin from start.

Since this is StudioAssist, I will help @duy.lam answer this.

@arvind.choudhary This is an item we have planned in our roadmap too, thanks for sharing this feedback with us - it’s really helpful for us. We’ll keep you updated once it’s released.

[xuan.tran]

@arvind.choudhary

Why isn’t Studio Assist able to automatically look into my project’s test artifacts and give context‑aware explanations? Since it is integrated with Katalon Studio, shouldn’t it detect variable declarations, test suite bindings, and global variable usage without me explaining them? For example, if I ask about a failure in Test Case X, shouldn’t it automatically analyze the test case, suite, and bindings to identify missing links or configuration issues?

@arvind.choudhary You can get the project context by selecting the current file or selecting files from your project. If you mentioned the failure in Test Case X, in the agent mode, I believe the agent will start getting test case details too. However, for data binding, this use case is not supported yet. Thanks for raising this idea with us. We’ll look into it further

[xuan.tran]

@arvind.choudhary

Will agentic workflows maintain an execution memory across multiple steps — allowing the agent to recall earlier failures, object patterns, or app state transitions to inform future decisions?

@arvind.choudhary Currently not. Is there any reason why you think adding this info to memory will help you in your QE tasks? - I’m interested to hear more

[xuan.tran]

@arvind.choudhary

As MCP servers extend StudioAssist’s capabilities, will future releases allow custom MCP server hooks for advanced tasks like live DOM introspection, codebase search, or external requirement mapping (e.g., Jira integration)?

@arvind.choudhary In KS 11, which we’ve just released, we added more tools that can get the page source and the accessibility tree from your web page, which makes test case and test object generation easy and more accurate.

Atlassian MCP server is supported since KS 10.4.0, too. For codebase search, currently we cannot connect to GitHub MCP yet, that’s for another future release then

[xuan.tran]

@arvind.choudhary

My another question is around Autonomous Data Binding Validation

Will StudioAssist’s Agent Mode eventually be able to verify data binding correctness across test suites — such as checking variable type mismatches, missing dataset bindings, or unused parameters — before execution?

This is a great idea. Currently, with StudioAssist, we can create test data and test suite already. We’re also planning to make data binding easier - so this validation idea is really helpful for us. We’ll keep you updated

[xuan.tran]

@arvind.choudhary

scan the entire Object Repository, detect duplicated test objects, unify similar locators, and refactor test references across the project automatically

Basically we can, now with Agent mode

[xuan.tran]

@arvind.choudhary

My another question is around Autonomous Test Healing

This is related to Self-healing, so @Shin and @duy.lam can help