e2e gets $8 million to test code that AI agents write
The tool does the steps of an AI agent again and uses the AI model only when the interface changes.
e2e, a startup that had the name TesterArmy, says that on November 3, 2026 it got $8 million from investors. It makes tools that check software that AI agents write. The open-source framework of e2e had 1.5 million downloads in its first month, e2e says. The framework is not at version 1.0 at this time.
How it works
Developers write tests in TypeScript for web, iOS and Android software through one API. They can add steps where an AI agent follows a task. When an agent step passes its check, e2e keeps the steps and does them again. It does not use the AI model for this. If a change to the interface breaks this, the live agent finds a new path.
The numbers
e2e says the framework had 1.5 million downloads, 12,000 GitHub stars and 60 external contributors in its first month. In one day, it had about 100,000 downloads. These numbers come from the company. It gave no test results for this method.
What is not known
The project text tells user that APIs and settings can change between small releases. The framework does not work with desktop software. The company will start a cloud tool in 2027 and gave no revenue numbers.
Insight: Checking agent code is becoming a market of its own
In e2e, the tests are code that the user has, and the user selects the AI model. iMerit says that experts find errors that automatic tests do not find. This reads as a move to control the check on agent work. Open code is on one side, experts on the other. The group that sets how teams check agent code can control a step of AI work.
Sources
Posted
