# How do I know my app works?

Checked 2026-11-15 against GilbApp d5296208bbffa2ab70076e6194f5537c4be1ddf7.

A separate agent tests each piece when it can; you try it yourself on your computer, and you check it once it is live.

## Deeper

Your app works when it does what you asked, for the people it is for, where they use it. In the way Kai-Zen works, only something that was actually run counts as proof, such as a test or a try in a browser. Your AI saying that it works is not proof, and neither are two agents agreeing.

Four checks come one after the other.

Tests by a separate agent. Agent.Coder, the agent that writes your app's code, has each part tested by Agent.Tester, the agent that tests what others built. An agent is your AI following one written procedure, its standing instructions, for one kind of work. Agent.Tester is started as a separate agent that knows nothing of how the part was built. A test is a small program that uses one part of your app and checks that the result is the one expected. Agent.Tester saves a report with its verdict: PASS, PARTIAL \(only some of the checks could be run\) or FAIL. If Agent.Coder cannot start a separate tester, it tests the part itself and marks the result self-QA, which may stand in for a PASS from Agent.Tester but is not equal to one. Agent.Scaling, the agent that checks what was built, checks it too.

A look in a real browser. For a change to a screen, Agent.Coder, if it has a browser tool, opens the screen in a real browser at phone size, takes a screenshot and counts the errors the browser reports. If it has none, it says so and the agent that handed the work to it takes that proof. A passing test run is not this proof.

Your own try. When the work you asked for is built and tested, Agent.Coder starts it on your computer and tells you what to try and what you should see. The answer [Where can I test what I've built, and try it myself before anyone else sees it?](http://support.gilb.com/help/where-can-i-test-what-ive-built-and-try-it-myself-before-anyone-else-sees-it) shows how.

Your check once it is live. Nothing goes live without your yes. Once it is live, Kai-Zen, the planning agent, asks you to check it live, fixes what you send back and asks again, until you approve. Then it has each goal the step aims at measured where the app's users use it, by an agent other than the one that built it; a goal here is a goal for how well your app must do something. A pass on the builder's own computer does not count as working in production, the live system people use.

Where you see the result: in your AI's reply when it hands work back. Each finished piece is marked "REAL" if it ran on the real thing, "MOCK" if it ran on a stand-in, or "UNPROVEN" if nothing was run; the answer [What counts as proof that something works?](http://support.gilb.com/help/what-counts-as-proof-that-something-works) explains them. In GilbApp's plan, the "Status", where something stands now, of a step, a Function \(something your app must do\) or a Solution \(a way to build something\) shows "Developed" only after the tester's saved PASS, and "Production" only once it is proved live.

An AI can be wrong. If something does not work for you although the tests passed, see [My AI says the tests passed, but it does not work for me. What now?](http://support.gilb.com/help/my-ai-says-the-tests-passed-but-it-does-not-work-for-me-what-now)

This answer is written for a computer screen.

Source: Standards read from GilbPlatform on 13 November 2026: Kai-Zen.EvidenceRules \(Agree: "Agents agreeing is not proof. Only a run is."; Tag: REAL, MOCK, UNPROVEN\), Agent.Coder \(AfterIncrement: separate Agent.Tester, RuleQatestIsolation, self-QA "may stand in for an Agent.Tester PASS but is not equal to one", phone-size browser proof with screenshot and console error count, Agent.Coder \(Browser: "If you have no browser tool, say so"\), "A passing test run is not this proof", after the last increment run Proc\_v\_i\_c\_Handback; Solutions: Developed only after the tester's saved PASS, Production only when proved live, "A pass on your own computer is not production"\), Proc\_v\_i\_c\_Handback \(start the dev server; "Go try it.", "Open:", "Test:"\), Agent.Tester \(Task: saved report, verdict per RuleQatestPartialIsLawful\), RuleQatestPartialIsLawful \(PASS, PARTIAL, FAIL\), Kai-Zen.Cycle \(Check, Live, Human, Measure\), Kai-Zen.HumanCheck \(Ask, Fix\), Kai-Zen.GoLive, Kai-Zen.ValueDeliveryStepRules \(Status, Delivered\). GilbApp 49b1ae53 \(https://gilb.com/app/version.json, read 13 November 2026\): src/lib/base/planCard.ts:31 \(Status words Planned, Building, Developed, Production, Retired\). Live help pages, sentences shared: What counts as proof that something works?; How do I build an app with an AI coder?; Can I trust what the AI suggests and builds?; How does Kai-Zen run a whole delivery cycle?
