Welcome to TestForge AI — and what Free for Launch means

Today we're opening TestForge AI to every team — for free, for as long as our launch runs. No credit card, no trial countdown, no feature gating. If you're a QA team anywhere shipping product to customers, you can sign up at testforge-ai.com and run your first regression suite against your real application before lunch.

This post is a short note on what TestForge is, who it's for, and what Free for Launch actually means.

What we built

TestForge is a platform that takes a written product requirement and turns it into a working, self-healing test suite — and uses an AI analyst to explain every failure in plain English.

You paste a requirement document. The platform drafts a complete suite of Gherkin scenarios for you to review. You approve, revise, or reject each one. Approved scenarios become real Playwright test files automatically. The page-object model is derived from your live application by scraping the DOM. Tests run in disposable Docker containers across Chromium and Firefox, capture per-step screenshots for visual regression, and produce a structured run report.

When something fails, an AI analyst — built on Anthropic Claude — reads the failure and writes a structured explanation: what was expected, what actually happened, the most likely causes ranked by probability, and a suggested fix. Real defects get a draft Jira ticket attached automatically.

When the UI shifts and your locators break, the platform tries to recover them at runtime. Successful heals are logged; the drift pattern is learned per project and surfaced so you can update the page-object permanently.

Who it's for

TestForge is built for QA and engineering teams who spend more time on the plumbing of test automation than on what testing is actually for: protecting your users from regressions. That includes:

• QA engineers who are tired of fighting Selenium locators after every deploy and triaging the same flaky failures every Monday.
• Software developers who want tests written from PRDs before the PR is even merged — and a four-line plain-English explanation when CI fails.
• Engineering managers who want a cleaner signal on what's really broken in the product, separated from the noise of flaky and environmental failures.
• CTOs and VP Engineering who want a release-confidence story they can actually defend, with governance, audit, and risk scoring baked in.

What Free for Launch means

Every team that signs up during our launch period gets full access to the platform — unlimited test generation, self-healing Playwright suites, per-step visual regression baselines, AI defect analysis with plain-English narratives, drafted Jira tickets, webhooks. No credit card. No countdown timer.

When your team starts hitting premium-model usage or higher run quotas, you'll upgrade to a paid plan with bigger quotas, the API surface, and an SLA. The work you've already done stays yours either way — your test scenarios, your visual baselines, your flywheel data are not held hostage to a tier upgrade.

We're doing it this way because the only way to know whether TestForge works for your team is to point it at your real requirements and your real application. A 30-second demo video doesn't answer that question; running it against your codebase does. Free for Launch is how we make that easy.

Why now

Three things converged this year that made TestForge possible.

Anthropic's Claude got good enough — and structured enough — to read a real failure and write a real explanation. Not just a generic stack-trace summary, but a four-section narrative that reads like it was written by a senior tester who already understood the codebase. We spent months tuning the prompt + governance + caching to make this fast, cheap, and reliable.

Playwright's API matured enough that we could safely scaffold tests from a live DOM, including handling per-step screenshots, trace files, and a healing wrapper around every locator. We didn't have to invent a new test runner; we could stand on a great one.

And test automation, as a category, finally had enough painful real-world failure modes — flaky tests, brittle locators, useless triage spreadsheets — that the appetite for a fundamentally better way is widely shared. Every QA engineer we've talked to in the last six months recognised the problem before we finished describing it.

Try it

Sign up at testforge-ai.com and run your first regression suite this afternoon. If you'd rather see the platform guided, click Book a Demo from the home page and we'll reach out within one business day.

Thanks for reading — and welcome to TestForge AI.