TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Momentic is presenting Mo, an AI quality-assurance agent that runs end-to-end tests against web apps and helps teams check changes in development and CI. The company says its platform has executed more than 70 million test runs, but the provided figures and customer results are company-reported; independent validation and detailed measurement methods are not included.
Momentic, a Y Combinator Winter 2024 company, is promoting Mo, an AI quality-assurance agent that tests applications by running user flows and checking expected outcomes. The announcement matters to software teams adopting coding agents because it addresses a growing development challenge: verifying that code changes work in the actual app before they are merged or deployed.
Momentic says teams can describe a flow and its expected result in plain English, then store the test steps and assertions in YAML files in their repository. The platform can run those tests locally or through a command-line interface in continuous integration, where a test result can serve as a required check before a merge or deployment.
The company describes features for supplying tests with product-specific guidance, inspecting screenshots and traces when a step fails, and distinguishing application bugs from test or environment problems. When a user interface changes, Momentic says Mo can try to find controls again during a run; proposed lasting changes to a test file are presented for review rather than applied automatically.
Momentic’s website reports 70,607,819 test runs, 339,166 changes verified, 80,290 pull requests verified and 117,010 bugs caught. These are the company’s own cumulative figures, and the supplied material does not state the reporting period or provide an independent audit. Customer testimonials also describe faster test writing and execution, but those accounts are not controlled comparisons.
Testing Moves Into the Coding Loop
AI coding tools can produce changes quickly, but software still needs to be checked against the behavior users rely on. Momentic’s pitch is to make end-to-end verification part of everyday development, rather than leave it to a separate manual testing stage or a later release review.
If the approach works reliably, teams could run repeatable checks on more changes and get evidence from the app itself before merging code. That may reduce the time between introducing a defect and finding it. The benefit depends on the tests accurately representing important user journeys and on teams being able to tell genuine product failures from flaky tests or problems in the test environment.
The product also illustrates a broader shift in how AI agents are used in software work: an agent can be asked not just to write code, but to interact with a running application and test an intended result. Momentic says its tests can be reviewed alongside code and run in CI, giving developers a familiar way to inspect and gate changes. The announcement does not establish how well the system performs across different applications or how its reported bug counts compare with other testing methods.
Products Worth Considering
Samsung Odyssey G4 25-inch FHD Gaming Monitor
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
Amazon Kindle Scribe (64GB) with Premium Pen, Metallic Jade
As an affiliate, we earn on qualifying purchases.
How Momentic Says Mo Works
Momentic frames its product around the claim that faster code production makes verification a development bottleneck. Its workflow starts with a description of a user journey and expected behavior. Tests can then run against the app under development, with results and failure evidence available for inspection.
The platform’s Knowledge Base can hold product terminology, business rules and expected behavior, which Momentic says AI steps consult while running. The company also describes an app graph built from screens and flows observed during tests, along with a process for reviewing suggested additions to product guidance.
For teams using coding agents, Momentic says an agent can use the Model Context Protocol (MCP) to author a test, check it in a live session and save the YAML file with the code. The source material also lists examples of bugs the company says its system detected, including a skipped new-device verification step and a mobile page-creation issue. No customer-by-customer verification or technical case-study detail was supplied for those examples.
“Setting up tests is simple and the developer experience has been fantastic.”
— Erdem, Head of Developer Experience at Notion, in a testimonial published by Momentic
Independent Performance Evidence Is Limited
The announcement does not specify when Momentic’s cumulative usage figures were measured, how the company defines a test run, a verified change or a bug caught, or whether those totals have been independently checked. It also does not provide a benchmark comparing Mo with conventional automated testing or other AI testing products.
Customer quotes and listed bug examples offer useful indications of intended use, but the supplied material does not include detailed methods for validating the results. It remains unclear how frequently the agent’s recovery and proposed test repairs succeed, how much human review they require, and how performance varies across application types. Pricing, availability limits and further technical details are also not stated in the source material.
Teams Can Test Through CI
Momentic’s described next step for teams is to write or generate tests, review them in the repository, and run them locally and in CI. The company says CI exit codes can be used as required checks before code is merged or deployed. That makes the product’s practical value dependent on how it behaves across a team’s own applications and workflows, not only on the platform-wide totals cited in its announcement.
Further evidence would include clear reporting periods and definitions for the company’s metrics, independently reviewable customer results, and information about test reliability and repair rates. The supplied announcement does not identify a future release date or a specific upcoming product milestone.
Key Questions
What is Momentic announcing?
Momentic is promoting Mo, an AI quality-assurance agent designed to run end-to-end tests against applications and help teams check changes before release.
How are tests written?
Momentic says users can describe a flow and its expected result in plain English. Test steps and assertions are stored in YAML files that developers and coding agents can review.
Can Momentic run tests before a merge?
Yes. The company says its CLI can run tests in CI, with an exit code used as a required check before a merge or deployment.
Are Momentic’s performance figures independently verified?
The supplied material gives company-reported totals, including more than 70 million test runs, but does not describe independent verification, a reporting period or the methods used to calculate the figures.
What happens when an app’s interface changes?
Momentic says Mo can look for controls again during a test run and propose lasting edits to test files for a person to review. The announcement does not provide an independent success rate for these repairs.
Source: momentic.ai
