What you can do with Testmode
Testmode’s features in one place: plain-English tests compiled into repeatable browser actions, Self-driving that explores your app and keeps the suite up to date, investigations of failed runs, findings, schedules, the assistant, CI and MCP.
Testmode is an AI testing tool for web apps. You describe tests in plain English, or let Self-driving write them; Testmode runs them in real cloud browsers and shows you exactly what happened.
Write tests in plain English
Section titled “Write tests in plain English”A test case is steps and an Expected Outcome, written the way you’d explain them to a colleague. No code, no selectors, no recorder. The first time a test runs, Testmode turns each step into browser operations and saves them, so every later run does the same thing, until you reword a step. See Write a test case.
Run them anywhere you deploy
Section titled “Run them anywhere you deploy”Point the same test cases at staging, production or a preview: environments hold the URLs, tests don’t. Tests of a run execute at the same time, each in its own fresh browser. Tests that change real data stay off production unless you allow them. See Environments and Run tests.
Start signed in
Section titled “Start signed in”Store a test login once, and tests start already signed in. Tests that share a login reuse one saved session instead of each signing in. See Credentials and automatic login.
See exactly what happened
Section titled “See exactly what happened”Every test has a verdict with a reason, a recording of the browser, the operations each step ran, a screenshot of the moment it failed, and the browser’s console and network logs. Editors can watch the browser live while a test runs.
See Read run results.
Let Self-driving test your app
Section titled “Let Self-driving test your app”Turn on Self-driving and Testmode tests your app on its own. It explores the app a little every day and keeps a map of its screens and journeys; it writes, fixes and schedules tests by your goal; and when a run fails, it re-runs the failures and says what each is: a bug, a flaky test, a test to update or an environment problem. Every change it makes comes with Undo, and what needs you arrives in the Inbox, or in Slack. See Turn on Self-driving.
Track what testing found
Section titled “Track what testing found”Every failure your app causes files a finding. Triage it as a bug, an intended change, a test problem, flaky or an environment problem, and follow a bug until a passing run verifies the fix. See Findings.
Run on a schedule, get reports
Section titled “Run on a schedule, get reports”Group test cases into plans, schedule them hourly, daily, weekly or monthly, and email the results to whoever needs them, or only when something fails. See Test plans, Schedule test plans and Email run reports.
Ask the assistant
Section titled “Ask the assistant”Open the assistant on any page and describe what to test: it drafts test cases, plans, schedules and variables, and saves them when you agree. It also knows the map and the Inbox, so you can ask what is untested or what a failure means. See Ask the assistant.
Plug into CI and AI agents
Section titled “Plug into CI and AI agents”Run tests from GitHub Actions or any pipeline with a project API key, tell Self-driving when you deploy, and let AI agents such as Claude, ChatGPT or Cursor list your tests and start runs over MCP. See Run tests from CI and AI agents.
Work as a team
Section titled “Work as a team”Invite your team with four roles, from Viewer to Owner, limit who can join to your email domains, and work across several organizations with one account. See Manage your organization.