The Turkish company Text2Test is developing an AI-powered software testing platform that lets users describe what a product should do in natural language and then converts that description into an executable test running on real browsers. The company says the process requires neither code writing nor the use of fragile interface selectors, and that the tests can keep pace with application changes.
Text2Test was founded in May 2024 by Esin Yılmaz, chief executive officer, and Prof. Dr. Cemal Yılmaz, chief technology officer, and its team currently consists of nine people. According to information provided by the team, the company developed its infrastructure in-house over approximately two years of research and development, rather than relying on a ready-made software testing platform.
From Selectors to Semantic Reading
Traditional interface-testing tools often rely on code or element selectors, which can make tests prone to failure when the interface structure changes. Text2Test, however, says its models read the screen semantically and visually, more like a human testing expert would understand it, rather than analyzing raw HTML or selectors alone. According to the company, the test is linked to what the product is supposed to do, not to the way the code was written.
The company positions its product within the AI-powered test automation category, which includes traditional selector-based tools, as well as no-code platforms and solutions built on large language models. Text2Test also highlights its support for an MCP server, which can be connected to requirements, design and code sources such as Jira, Figma, GitHub and OpenAPI, with the aim of generating test suites from preexisting documents.
What Changes in Practice?
Text2Test says token consumption in AI-powered testing tools typically occurs during every run, whereas tokens on its platform are consumed while the test is being assembled. As a result, according to the company’s description, the cost factor is tied to creating tests rather than to the number of times they are run. The platform currently supports testing web applications. The company says native iOS and Android application testing is also available, with the ability to write a test once and run it on both platforms, but it is working to make this function available through its self-service model.
Funding, Customers and Expansion
Text2Test announced that it raised $495,000 in a bridge funding round, noting that the round has not yet been completed. The platform operates under a subscription-based SaaS model, while its early customers came from the banking and insurance sectors. Because of the detailed regulatory requirements in these two sectors, the revenue model began with deployment options within the customer’s local infrastructure, alongside enterprise packages that include service-level agreements and customized contracts.
The company has also launched its self-service model with monthly subscription plans targeting different volumes and needs, and says it wants to reach companies in Turkey, Europe, Asia and America through it. Its next plan is to expand mobile application testing through a local runner that connects a Mac computer and the user’s devices or emulators.
This step reveals a practical direction in test automation: reducing reliance on manually writing and maintaining tests and linking them directly to requirements. However, most of the platform’s claims, such as tests remaining unaffected by interface changes and lower operating costs, remain statements from the company itself in this article and do not include independent figures on success rates or performance comparisons with competing tools.