[Katalon Studio] Turn Your Tests into Automation with AI WebUI

[nghi.phan]

BEST PRACTICES · AI & PRODUCTIVITY · Katalon Studio 11.5 · Web UI testing · September 2026

Got manual test cases in a file, or user stories in Jira? This guide shows how the Katalon AI Assistant turns them into runnable web tests — and how you stay in control.

1. Is this you?

You have a folder full of manual test cases, but only a few are automated.
Your user stories in Jira have clear acceptance criteria — yet tests come last, or never.
Writing test objects and scripts by hand eats most of your sprint.
You tried AI for code snippets, but never let it build a whole test.

2. The short answer

If you only read one part, read this.

• Start from what you have. A plain test case file or a Jira story is enough. No need to rewrite anything first.
• AI does the typing. It creates the test objects, the scripts, the variables — and even the test suite.
• You stay the reviewer. Every new file is listed for you to check, keep or undo.

3. How it works

You describe the goal once. The AI Assistant, in Agent mode, does the rest step by step — and shows you each step.

What you have: a test case file or a Jira story
↓
AI Assistant: reads it and plans the work
↓
New test assets: objects · test cases · test suite
↓
You review: keep what’s right, undo what’s not

The AI does the typing. You make the calls.

4. Pick your starting point

Both cases use the same demo website, so you can compare them side by side.

• Case A: “Our test cases live in a file.” Attach the file and ask the AI to automate it → Section 5
• Case B: “Our requirements live in Jira.” Connect Jira once, then give the AI the story link → Section 6

5. Case A — From a test case file to automated tests

Your team already wrote the steps, the test data and the expected results. The AI reads them and writes the automation.

The starting point: a plain text file of manual test cases — steps, data and expected results, as your team wrote them.

1. Open the Katalon AI Assistant and make sure Agent mode is selected.
2. Attach your file with the paperclip icon, or drag it into the chat.
3. Paste the prompt below and send it. Start with a few test cases, not all of them.
4. Watch the agent work. It reads the file, creates the test objects, then writes each test case.
5. Review the new files before you keep them (see Section 7).

Prompt used in the video:

Generate the first 3 test cases in the attached txt file.
Generate test scripts and test objects for those test cases.
Put them into the folder named "TXT File" created automatically by yourself.

One file, one prompt (left). Every new file is listed, ready for your review (right).

What you get: 3 manual test cases became 3 runnable test cases plus their test objects — 32 new files in one conversation, all waiting for your review.

Tip: go in batches of three to five. Check the first batch, then ask for the next one in the same chat — the AI keeps the context and reuses the objects it already made.

Watch: Generate test cases from a TXT file (about 2 minutes)

Generate Test Cases From a TXT File with Katalon StudioAssist

6. Case B — From a Jira story to a ready test suite

Your acceptance criteria already say what “done” looks like. The AI turns each one into a test.

The starting point: a Jira story with acceptance criteria (AC1, AC2 …).

First time only: connect Jira (about a minute)

1. In the AI Assistant, open Agent Mode Setting → MCP Tool → Add MCP Server.
2. Choose HTTP, enter the server URL https://mcp.atlassian.com/v1/mcp, give it a name such as atlassian, and click OK.
3. Sign in to Atlassian in the browser window that opens, pick the apps you allow (for example Jira), and click Accept.

Add the Atlassian server once (left). You see exactly what access you are giving (right).

Then: turn the story into tests

1. Paste the story link into the prompt below and send it.
2. Approve the Jira tools when asked — Allow Once, Allow this Conversation or Always Allow. Nothing runs without your OK.
3. Let it build. It creates shared test objects, then one test case per acceptance criterion.
4. Ask for a test suite with the second prompt, so the whole story runs together.

Prompts used in the video:

Create test cases for this user story <your Jira story link>.
Each AC in the requirement should be a separate group of test cases.
Generate scripts and test objects for those test cases.
Put all generated files into a folder named "User Story JIRA".

Create a new test suite named "User Story JIRA Test Suite"
and put all generated test cases to that test suite.

The agent explains its plan and asks before it uses a tool (left). One test case per acceptance criterion, plus shared objects (middle). A plain-language summary of what was created — then one more prompt for the test suite (right).

What you get: one story with 6 acceptance criteria became 16 shared test objects, a test case for each criterion, and a test suite to run them all.

Tip: “one group per AC” keeps every test traceable to the requirement. When a test fails, you know exactly which criterion broke.

Watch: Generate test cases from a Jira ticket (about 5 minutes)

Generate Test Cases From a JIRA Ticket with Katalon StudioAssist

7. Review it like a teammate’s pull request

AI is fast, but it can misread a step. Five minutes of review turns a good draft into a test you can trust.

1. Open each new file from the list in the chat. New files are marked New; changed files open in a diff view that highlights the changes.
2. Check the locators. Open a few test objects and make sure they point at the right element.
3. Check the test data. Values like the URL, email and password should be variables, not typed into the script.
4. Run the tests once. Fix or undo anything that doesn’t match what the user really sees.
5. Finish the review. Undo single files, Undo All, or Mark All Reviewed when you’re happy.

Good sign: the AI put the URL, email and password into variables, so you can change them without touching the script.

8. Tips that work for both cases

• Say where to save. Name the folder in your prompt, so new files don’t mix with old ones.
• Ask for variables. Test data in variables is easier to change and to reuse with other data.
• Let tests share objects. When one locator changes later, you fix it once and every test that uses it is fixed too.
• Stay in one conversation. The AI remembers what it built, so follow-up requests stay consistent.
• Finish with a test suite. Group the new tests so they run together — and later, on a schedule.

9. Myths, cleared up

Myth: “AI-generated tests are ready to ship.”
Fact: Treat them as a strong first draft. Review them like a teammate’s work, then run them.

Myth: “We’d have to rewrite our manual cases in a special format first.”
Fact: In the video, the AI read a plain text file exactly as the team wrote it.

Myth: “Connecting Jira gives the AI free access to everything.”
Fact: You choose which apps to allow when you sign in, and you approve each tool before it runs.

10. Get started

• Katalon Studio 11.5 with the Katalon AI Assistant in Agent mode.
• A Katalon Studio Enterprise license to use Katalon AI out of the box, or your own AI provider key.
• For Case B: an Atlassian account that can open the Jira project.

Topic	Katalon Docs
Agent mode and prompt examples	Katalon AI Assistant Agent mode
AI Assistant features and licensing	Katalon AI Assistant Overview
Connecting external tools like Jira	Add an external MCP server
How AI is controlled in your account	Best practice guide: AI in Katalon Studio: how it works, and how you stay in control

Katalon — Best Practices Guide. Applies to Katalon Studio 11.5. Screens may vary by version. Screenshots from Katalon demo videos using a public demo website. Questions not covered here? Contact the Katalon Onboarding & Adoption team at onboarding@katalon.com.