Skip to content
TestAutomate Join the waitlistWaitlist

Get notified at launch

TestAutomate isn't released yet. Leave your name and email and we'll notify you when it launches.

Home / About COMPANY

We got tired of test suites nobody believes

TestAutomate is a small founding team building one thing: a QA agent whose red result means a real bug. This page is our letter to you.

A letter from the founding team · Aug 2026

Every engineering team we have ever worked on had the same artifact sitting in its repo: an abandoned test suite. It was written with real care, usually right after a bad outage or a big launch. For a few months it ran green and everyone felt safe. Then a designer renamed a button. A div moved. A selector broke. The suite went red for a reason that had nothing to do with the product.

Someone fixed the selector. Then three more broke. Then someone marked the flaky ones as skipped "for now". Then the suite went red on a Friday and the team shipped anyway, because everyone quietly knew the red was noise. That was the day the suite died. It kept running in CI for another year, but nobody looked at it again. QA went back to being a person clicking through the app before release — the exact thing the suite was built to end.

We built TestAutomate because that failure is not a discipline problem. It is a design problem. So we start from intent: you describe the test in plain English, an agent drives a real browser to carry it out, and an AI verifier grades the run with screenshot evidence. When your UI changes but the workflow still works, the test still passes — and tells you what changed.

We are pre-launch. We have no customer logos to show you, no testimonial carousel, no invented star rating. What we have is a working product we run on real apps every day, an honest changelog, and a strong opinion about what QA tooling should cost and where your data should live. If that resonates, keep reading.

WHY WE STARTED

Devi Lavanya founded TestAutomate after two decades in software quality assurance, watching the same pattern repeat across every era of the industry. In her enterprise-services years, batch-dependent test plans failed together, with a single upstream failure invalidating a full day of downstream runs. On a VOIP product, test failures regularly had nothing to do with the application, and separating a genuine defect from a network fault became a discipline of its own. In her Selenium years, UI redesigns broke locators at scale, and the effort of repairing them competed directly with the work of testing. Release weeks followed the same course everywhere: manual regression cycles ran long, and coverage was cut when time ran out.

The conclusion she drew from those years is the premise of this company. Most of what the industry calls test automation is maintenance of the automation itself, not testing of the product. The tooling encodes how to locate elements, and nothing about what actually matters.

Shravin Vivek, TestAutomate's Founding Engineer, turned that diagnosis into a product, and built it end to end. The browser agent reads a page and locates elements the way a person would, with no selectors to maintain. The verifier, a separate model, evaluates every run against written expectations rather than accepting the agent's own account of what happened. An ingestion pipeline converts product documentation into an executable regression suite. And the guardrails keep results trustworthy: a run whose preconditions failed is reported as blocked rather than failed, and no bug reaches an issue tracker without human review.

That division of responsibility still defines the company. The product is shaped by the problems Devi Lavanya spent her career confronting, and engineered by Shravin Vivek to remove them.

Who we are

Founder

Devi Lavanya

Devi Lavanya is the Founder of TestAutomate. She has been testing software since 2006, across enterprise services, a VOIP product, and health-tech platforms, She started TestAutomate to remove the maintenance burden that testing tools keep creating. Her posts draw on the problems she's run into doing that work.

Connect with Devi on LinkedIn ↗ All posts by Devi →

Recent: Automated UAT Testing: What It Really Takes · Best Agentic QA Tools in 2026: The New Category, Mapped · Best Test Case Generation Tools in 2026

Founding Engineer

Shravin Vivek

Shravin Vivek is the Founding Engineer of TestAutomate and built it end to end with Devi Lavanya: the browser agent, the verifier that grades every run, and the pipeline that turns product docs into regression suites. His posts explain the product's design decisions from the inside.

Connect with Shravin on LinkedIn ↗ All posts by Shravin →

Recent: Real Browser Testing vs Cloud Browsers · The Best AI Testing Tools in 2026, Compared Honestly · AI Testing Cost: Engineering the Bill Down

WHAT WE BELIEVE

Five beliefs, in the order we hold them

1

Intent over selectors

A test should say what a user is trying to do, not which CSS selector to click. Intent survives redesigns. Selectors don't.

2

A red result must mean a real bug

Every false alarm spends trust you can't buy back. Escalate-on-fail, blocked-is-not-failed, and flake labeling exist so that red stays meaningful.

3

Your key, your data, your infra

Bring your own AI key and pay model costs at cost. Run the whole stack on your own servers if you want. We are not in the data-hoarding business.

4

Evidence over assertions

"Assertion failed on line 40" convinces nobody. A verdict with screenshots of what the agent saw convinces everyone, including the engineer who has to fix it.

5

Tools compound, services don't

Every run teaches the platform your app — login recipes, nav quirks, playbooks. A managed QA service resets to zero with every new tester. A tool gets smarter.

DESIGN-PARTNER PROGRAM

Ten teams, before everyone else

We are onboarding ten design partners ahead of launch. It is deliberately small because we do the setup with you, not to you.

  • White-glove setup. We sit with your team, ingest your docs, and write your first suite together.
  • Launch pricing, locked. Whatever you pay as a design partner is your price. It does not go up when we launch.
  • A direct line. You talk to the people who wrote the code. Your bugs and requests shape the roadmap.
10design-partner spots
1:1founder-led onboarding
0fake testimonials on this site
We publish what's live and what's coming, and we mark the difference on every page. If a claim on this site turns out wrong, tell us and we'll fix the page.

Help us build the QA agent you'd actually trust

Ten spots. White-glove setup. Launch pricing locked for good.