Why traditional QA is no longer enough
In the era of AI programming, where tools like Claude Code, Cursor, or GitHub Copilot churn out hundreds of lines of code daily, traditional testing is no longer sufficient. A Faros AI study from 2026 shows that the monthly number of production incidents has increased by almost 58% since developers started widely using AI assistants. And a May survey by CloudBees among enterprise companies revealed that 81% of technology leaders confirmed a direct increase in production errors caused by AI-generated code.
In other words — code is being created faster than ever before, but testing, which has always been a bottleneck in development, is now falling even further behind. And this is exactly where Momentic comes in.
What is Momentic and who is behind it
Momentic is an agent-based platform for end-to-end testing of web and mobile applications. It was founded by Wei-Wei Wu and Jeff An, former engineers from companies like Robinhood, Qualtrics, WeWork, and Retool. The startup went through Y Combinator and in 2025 raised $15 million in Series A funding led by Standard Capital. The platform is built in San Francisco and is currently used by teams from companies such as Notion, Quora, Webflow, Runway, Xero, and Retool.
What makes Momentic different? It's not just another tool for writing test scripts. It's an autonomous AI system that learns your product, understands its terminology, knows which parts of the application are fragile, and writes, runs, and fixes tests itself. You only review the output. As one of their customers aptly put it: "The best testers know the product better than anyone else. They know every nook and cranny. And that's exactly what's missing in agent code."
How the platform works under the hood
Momentic uses YAML files written in natural English — no XPath selectors, no fragile CSS locators. Tests are readable by both humans and AI agents. The entire system is built on five pillars:
1. Knowledge base and memory
The platform learns from documentation, guides, codebase, and ticketing tools like Jira or Linear. It understands your product's terminology, knows what a bug is and what is an intentional change. The more data you feed into the system, the smarter it gets.
2. Explore Agent — coverage that grows with code
Every time a developer merges a pull request, the Explore Agent reads the diff, identifies changes, and suggests new or modified tests. It automatically maps new functionalities, renamed components, and edge cases that were missed in the previous sprint.
3. Failure classification with 96% accuracy
False positive tests (so-called "flaky" tests) are one of the worst things that can happen to your test suite — not because they are annoying, but because they teach developers to ignore errors. Momentic has its own agent for failure classification that distinguishes a real bug from an intentional change in the application, environment errors, or transient instability. If it's an intentional change, Momentic immediately opens a pull request with the test fix. If it's a real bug, the team receives an alert with full context. The result is a 96% signal-to-noise ratio.
4. Auto-healing — tests that fix themselves
When your application's UI changes, Momentic automatically recognizes that it's an intentional change (and not a bug) and fixes the test itself. To date, the platform has performed over 8.9 million automatic fixes (auto-heals).
5. CI integration with one command
Just run npx @momentic/wizard@latest and in no time you'll have your test suite integrated into any CI pipeline that can run Node.js. Tests run on hosted browsers, Android emulators, and iOS simulators — in parallel.
Numbers that speak for themselves
Momentic today records impressive statistics across all customers:
- 77,813 tests created
- 70.6 million test runs performed
- 8.9 million automatic test fixes
- 117,010 bugs detected before they reached production
- 80,290 pull requests verified
- 339,166 code changes verified
Specific results from customers are no less impressive: Quora reduced daily testing from 7 hours to 30 minutes and replaced over 500 manual test scenarios. Retool increased release frequency eightfold and saved 40 engineering hours per month. GPTZero accelerated release cycles by 80% and reduced the number of bugs leaked into production by 89%.
How much Momentic costs
The platform offers three pricing tiers, with the price based on actual usage — not on the number of developers in the team:
- Free — free forever, 2,000 credits per month (about 200 test runs), all basic features including AI testing. No credit card required.
- Pay-as-you-go — $125 per month + additional usage. 10,000 credits (about 1,000 runs), 2 Android + 1 iOS device, 5 phone numbers for SMS/OTP testing, shared Slack channel. Exceeding the limit costs $0.01875 per credit.
- Enterprise — custom pricing, SAML SSO, SCIM provisioning, audit log, unlimited mobile test duration, SLA, dedicated account manager.
For Czech developers and small teams, it's crucial that the Free plan does not require a credit card and is free forever — ideal for trying it out on smaller projects.
What this means for Czech developers and companies
While Momentic is in English and Czech localization is not yet available, this is not an obstacle — tests are written in simple English and the CLI tool works from anywhere. For Czech startups and software companies that are increasingly using AI assistants for development (whether it's GitHub Copilot, Cursor, or Claude Code), Momentic offers a solution to a growing problem: how to maintain code quality when machines generate it.
In the European context, it is also significant that Momentic has SOC 2 Type 2 certification — which is important for companies that must comply with strict data security requirements under European legislation.
Why now
The launch of the new platform comes at a time when the AI coding market is overheating. Tools like Claude Code, Cursor, OpenAI Codex, and dozens of others generate code faster than ever before — but no one is properly addressing what happens when that code reaches production. As data from Faros AI shows, incidents are skyrocketing and traditional QA teams can't keep up. Momentic is the first platform to systematically address this problem — agent-based AI that tests code from agent-based AI.
Wei-Wei Wu, CEO of Momentic, summarized it clearly: "Quality used to be a function of the size of your QA team. Every developer, on every team, regardless of size, deserves quality. And it should be easy, fun, and built into the foundations of how software is created."
Does Momentic support Czech or have localization for the Czech market?
Momentic is currently in English, and Czech localization is not yet available. However, tests are written in simple English, which is understandable for most developers. The CLI tool and all documentation are in English. The platform is globally available, and the Free plan does not require a credit card, so any Czech developer can try it out.
How does Momentic differ from traditional testing tools like Selenium or Cypress?
The fundamental difference is autonomy. While Selenium or Cypress require you to manually write tests and then maintain them with every UI change, Momentic uses AI agents that create, run, and fix tests themselves. You don't write selectors or XPath; you describe tests in plain English. Furthermore, the platform learns your product and automatically suggests new tests to cover gaps with every code change.
What happens if an AI agent in Momentic fixes a test incorrectly?
Momentic has a built-in failure classification agent that distinguishes between real bugs, intentional changes, and transient errors. When the platform suggests a test fix (auto-heal), the developer reviews it before merging — the fix comes as a pull request, not an automatic commit. The platform achieves a 96% signal-to-noise ratio, so false fixes are minimal.
