UIlicious

UIlicious is a Singapore-founded company that automates user-interface testing for web applications. Its products combine readable test scripts with AI-assisted authoring. On February 26, 2026, it added Auto-Pilot to TAMI Studio: an agent that explores an application and executes an instructed scenario, producing a script that users can review, save and replay.[5][4]

Tai Shi Ling’s AI Test Authoring announcement, published March 8, 2023 and updated September 8, 2025, describes TAMI 1.5 following a TAMI 1.0 beta on UI-licious Snippets. In the Web IDE, users press F2 to activate the assistant and describe a scenario such as “Login to Netflix, and change password”. TAMI generates a test script; users can also select existing script sections for extension or refactoring.[3]

The February 2026 Auto-Pilot release extended this workflow from generating scripts to exploring and acting within the application. After a run, the generated script can become a saved, replayable test case. On May 13, 2026, UIlicious added messages for steering an agent while it runs, plus pause and resume controls.[4]

As described on the company’s product page on September 18, 2026, TAMI Studio is the recommended platform for new users, combining test-case management with an assistant that analyses screenshots and recordings. The company describes TAMI suggesting cases, documenting procedures and expected outcomes, generating automation scripts, suggesting fixes and writing bug reports. The classic Web IDE remains an experience for existing customers, with browser-based test authoring, scheduling, notifications and basic AI functionality.[2]

The underlying low-code framework runs on JavaScript. Commands such as I.fill("Search", "Mount Fuji") and I.see("Wikipedia") express user actions and checks, while variables, loops, conditionals and functions support more complex tests. Reusable flows and datasets allow scripts to run with different inputs, including external JSON or CSV data. UIlicious provides remote parallel execution and interactive reports with step-by-step replay, screenshots and logs.[2]

These AI features followed an earlier low-code testing product. UIlicious was founded in 2016 by Shi Ling Tai and Eugene Cheah; March 2021 reporting identified Tai as CEO and Cheah as CTO.[5]

In a Singapore Global Network profile, first published in October 2021 and updated in February 2024, Tai recalled encountering inadequate testing processes while building interfaces at advertising-technology startup Knorex. She developed her idea for testing accessible to non-programmers in Entrepreneur First’s pioneer Asia cohort, where she met Cheah. She described their engineering responsibilities as complementary: she built the test engine, while Cheah built the supporting infrastructure.[7]

Tai recalled that Entrepreneur First required the founders to seek paid contracts before building the product. The profile reports five contracts, each promising $500 for a prototype; it does not specify the currency. The first year went into refining the prototype into a minimum sellable product, followed by work on pricing and marketing. The profile reports that reaching profitability took three and a half years. Tai also recalled months when the founders took no salary to keep servers running and employees paid.[7]

The profile describes Daimler becoming a major client after its engineers encountered the founders at a Singapore convention. Tai recalled that securing the customer helped subsequent investor conversations. March 2021 reporting named Daimler, Jones Lang LaSalle and recruitment platform Glints as clients.[7][5]

In her May 2, 2019 engineering article, Tai explained how UIlicious supported commands such as I.click("Login") without explicit CSS or XPath selectors. The engine analysed the DOM using semantics, accessibility attributes and common interface patterns to determine the intended element. That analysis involved flattening the DOM tree through array merges.[8]

Tai reported that a customer’s enlarged, deeply nested DOM exposed repeated Array.concat operations as a bottleneck. Commands normally taking roughly one second exceeded 40 seconds, and suites normally finishing in under 20 minutes took hours. Replacing the array-merging approach with push resolved the slowdown, according to her account.[8]

The article’s headline 945× ratio came from a separate Chrome microbenchmark merging size-10 integer arrays 10,000 times: Tai reported 0.40 operations per second for concat and 378 for push. It was a workload-specific comparison, not a product-wide speedup. The operations also differ: push mutates the first array, whereas concat creates a new one. Tai credited GPU.js co-creator @picocreator with discussing V8 behaviour and lending the benchmarking PC.[8]

The targeting engine continued to receive changes alongside the AI products. On June 23, 2026, UIlicious changed I.fill to select existing field contents with Ctrl/Cmd+A before entering replacement text, addressing cases where clearing a field restored its original value. The same release rebalanced element-matching attribute weights to prioritise attributes with higher intentionality.[4]

In March 2021, UIlicious announced US$1.5 million in pre-Series A funding, led by Monk’s Hill Ventures. The announced use of the financing was to expand product development, sales and marketing teams.[5][6]

In Digital News Asia’s March 25, 2021 report, Glints CTO Seah Ying Cong said UIlicious had cut the company’s manual QA run work by 80% and eliminated QA bottlenecks in its development cycle. He highlighted collaborative tests written in almost plain English, allowing stakeholders to understand acceptance criteria. The reduction was Glints’ reported experience.[6]