IT Tech Pulse Exclusive Interview with Mudit Singh, Co-Founder and Head of Growth, TestMu AI
Stay updated with us
Sign up for our newsletter
Mudit Singh, Co-Founder and Head of Growth at TestMu AI, shares in an IT Tech Pulse interview how agentic AI is transforming software testing, quality engineering, and digital trust in the era of AI-generated code.
AI is helping teams ship products faster than ever – how should software testing evolve to stop speed from creating costly production failures?
We’ve entered what we call the era of Infinite Code. Cycles that used to take weeks now collapse into hours, and the volume of AI-generated code has overwhelmed every test stack built on yesterday’s assumptions. The economics make the case on their own: a bug caught in development costs 30 to 100 times less than the same bug surfacing in production. Testing has to evolve in the same direction the code is moving, autonomous, agent-driven, and embedded directly into the development loop. That’s the philosophy behind TestMu AI’s Kane CLI, a single terminal command that lets developers and AI coding agents run real browser checks before code moves toward production. Across our platform, we now run more than 1.5 billion tests a year, and the pattern is consistent: when verification keeps pace with code generation, production failures stop being inevitable and start becoming preventable.
Read More: ITTech Pulse Exclusive Interview with Rosalyn Curato, Chief Innovation Officer and GM of Agentic Security at Vouched
What truly separates agentic AI testing from traditional automation frameworks, and why are enterprises now giving this space serious attention?
It’s the difference between a co-pilot and an autopilot. Traditional automation, even when wrapped in LLM calls, is script-bound and reactive. Agentic AI testing is architecturally different.
We define agentic through four capabilities our system demonstrates end to end. It reasons over context, analysing code changes, historical failures, and risk signals to decide what should be tested and why. It acts autonomously, generating end-to-end tests from natural language and adapting them as the application evolves. It observes and learns, separating environment noise from genuine regressions and offering specific remediation. And it collaborates across roles — Planning, Authoring, Execution, and Analysis agents working as one coordinated system rather than bolt-on features.
Enterprises are paying attention because at the speed AI now produces code, deterministic scripts simply cannot keep up. Customers like Microsoft, NVIDIA, Bajaj Finserv, and Airtel have arrived at the same conclusion: only intelligent agents working alongside humans can sustain quality at the velocity of modern development.
Many organizations are rapidly adopting GenAI tools – what major quality, compliance, or security risks are still being underestimated today?
The biggest blind spot is the trust deficit in AI-generated code. With “vibe coding” now mainstream, large portions of production code are written by AI and never reviewed line by line — introducing silent vulnerabilities, inconsistent business logic, and edge cases nobody has thought through. The second underestimated risk is hallucination inside the testing process itself. An agent that is confidently wrong erodes trust in the entire quality pipeline.
We address this with three structural safeguards. First, full audit trails — every Analysis Agent decision shows its reasoning, so a human can validate the logic in seconds rather than guessing. Second, what we call the Strategic Human loop: agents handle the infinite-code volume, humans remain the arbiters on high-risk business logic. The AI proposes; the human disposes. Third, phased autonomy — we advise enterprises to onboard AI agents the way they would new hires, starting on low-risk regression suites and expanding scope only as reliability is proven.
The broader concerns around data privacy, bias, and explainability are real too. We mitigate the data dimension by running tests in development and staging using synthetic data, so AI experimentation never touches production data. That’s one reason regulated sectors like BFSI and healthcare have leaned in.
How is TestMu AI helping engineering teams confidently test AI agents, chatbots, copilots, and voice-based digital experiences at scale?
Testing AI products or digital experiences is fundamentally different from testing deterministic software, because the same input can produce different outputs. The TestMu AI platform is purpose-built for that reality.
Through Kane AI and Kane CLI, teams validate agentic and conversational systems using natural-language test definitions, intent-based verification, and automated bug discovery. The Planning Agent assesses risk and decides what to test; the Authoring Agent generates the cases; the Execution Agent runs them across thousands of browser, OS, and device combinations on our real device cloud; the Analysis Agent triages results and surfaces specific fixes.
For multimodal experiences, voice copilots, chatbots, mobile assistants, that means continuous validation at a scale no human QA team can match. Roughly half of our revenue today comes from AI-led products, and that depth of usage is what hardens the platform against the messy real-world edge cases AI agents actually encounter in production.
Read More: ITTech Pulse Exclusive Interview with Michael Jacobs, Head of Social Innovation at IBM
There’s growing debate around AI replacing jobs – will GenAI reduce QA roles, or create smarter and more strategic quality careers?
Agentic QA is a force multiplier, not a replacement. The biggest mistake leadership can make is hearing “autonomous testing” and assuming it means cutting QA headcount — that is a recipe for disaster. What GenAI eliminates is the maintenance tax: the brittle scripts, the flaky-test triage, the manual regression cycles that today consume up to 40% of QA capacity. What it elevates is everything else — designing test strategies for non-deterministic systems, defining success states, auditing AI outputs for bias and hallucinations, exploratory testing, and owning the quality narrative across the SDLC. We built TestMu AI to give quality engineers superpowers. The companies that will win are the ones investing equally in upskilling — letting AI handle scale and speed while humans bring judgement and context. With Kane AI, we’ve already seen quality democratised further: product managers, designers, and business analysts can now participate in testing without writing code, while QA engineers move up the value chain into quality leadership.
In 2026, which testing KPIs matter most for modern businesses: release speed, reliability, coverage, hallucination control, or customer experience?
Release speed is now table stakes, everyone is fast. The differentiators in 2026 are reliability, customer experience, and hallucination control. But the more interesting shift is the new set of metrics that emerge specifically for agentic quality engineering, this includes:
- Mean Time to Quality: how quickly a change moves from commit to validated, production-ready confidence.
- Autonomous Resolution Rate: the percentage of test failures triaged and resolved by agents
- Maintenance Burden Reduction: how much human effort is freed from supervising tests and redirected into genuine quality engineering.
Together with traditional reliability and CX metrics, these define what good looks like in an AI-first delivery model. The headline shift is from “how fast did we ship” to “how trustworthy is what we shipped” and that’s the metric boards and CTOs are increasingly aligned around.
Startups want faster launches with smaller teams and tighter budgets – what testing stack or framework should they prioritize first, and why?
We’re in a moment where building and shipping products has never been easier. With vibe coding and agentic development tools, even a 2-member team can ship. Smaller teams, tighter budgets, and faster launches aren’t aspirational anymore, they’re becoming the default. But that same leverage is exactly why testing can’t be an afterthought: when AI is generating the majority of your codebase, the volume of un-verified logic shipping into production grows just as fast as your velocity does.
Startups should resist the temptation to defer testing until after product-market fit. That debt compounds painfully, a production bug costs 30 to 100 times more to fix than one caught in development.
The right starting point is a lightweight, AI-native quality engineering platform that integrates into the development loop from day one.
To give an example of how TestMu can help, let’s look at Kane CLI, which is built to help you author tests from your CLI: it installs in minutes via npm or Homebrew, works seamlessly with AI coding agents like Claude Code, Codex CLI, Cursor, and Gemini CLI, runs real browser checks from the terminal, and slots into CI/CD without heavy setup. We also offer a free entry tier so smaller teams can start with zero financial friction.
The principle is simple: every feature shipped from a prompt is a feature nobody has actually verified, and at startup speed, “we’ll click through it later” becomes a liability that compounds quickly. Whether a team is on Selenium, Playwright, Cypress, or Appium, TestMu AI platform makes those frameworks instantly more powerful by handling the maintenance and scalability problems that typically might break in-house builds.
Looking ahead, ho/w do you see GenAI reshaping software delivery, developer workflows, and digital trust across the tech industry over the next five years?
Over the next five years, GenAI will absorb some specific areas in the software delivery pipeline like code generation, test authoring, deployment orchestration, even incident triage. Developer workflows will become conversational and intent-driven, with humans setting direction and AI agents doing the execution. Pricing models will shift in parallel, from per-seat licences to capacity-based models priced on agent usage, because the unit of value is moving from human logins to autonomous agent capacity.
The total addressable market for testing will expand dramatically, as code becomes infinite, the surface area to verify grows in proportion.
The deepest challenge of this era, though, isn’t tooling. It’s quality ownership. The defining work of the next five years will be embedding quality earlier in the lifecycle, surfacing risk signals at the commit level, and treating digital trust, transparency, explainability, fairness, continuous verification, as a product feature, not a compliance afterthought. That’s the future TestMu AI is built for.
Thank you Mudit, for taking the time to share your insights with us.
Write to us [wasim.a@demandmediaagency.com] to learn more about our exclusive editorial packages and programmes.
Mudit Singh is the Co-Founder and Head of Growth at TestMu AI (formerly LambdaTest), the world’s first full-stack Agentic AI Quality Engineering platform. A seasoned technology marketing and growth leader, Mudit has spent over a decade building go-to-market strategies, developer communities, and category-defining brands at the intersection of software quality and artificial intelligence.
Joining LambdaTest as Head of Marketing and Growth, Mudit played a pivotal role in scaling the company from an emerging cloud testing player into a globally recognised quality engineering platform trusted by 2.8 million developers and testers across more than 90 countries. He led the growth engine that powered the company’s expansion to 18,000+ enterprise customers, including Microsoft, OpenAI, NVIDIA, and Vimeo.
His work sits at the forefront of a fundamental shift in how software is built, tested, and shipped – championing a future where AI agents autonomously plan, author, execute, and analyse software quality so that human ingenuity can focus on what matters most: innovation.
TestMu AI is the world’s first Agentic AI-powered Quality Engineering platform, designed to enable organizations to automate and scale testing with intelligence at its core. Founded in 2017 and headquartered in San Francisco, the company serves over 2.8 million developers and testers across 18,000+ enterprises in 90+ countries.
Co-founded by Asad Khan (CEO), Jay Singh (CCO), Mayank Bhola (Head of Product), and Mudit Singh (Head of Growth), TestMu AI was built to close the gap between AI-accelerated code generation and traditional testing bottlenecks. By combining autonomous capabilities with seamless integration across modern development workflows, the platform empowers teams to deliver faster, more reliable, and secure software in an AI-first world.
Its AI-native product suite includes Kane AI — a GenAI-powered test agent that generates and executes tests from natural language — alongside HyperExecute for next-gen test orchestration, a Real Device Cloud, cross-browser testing across 3,000+ environments, and a Visual Testing Cloud. The platform powers 1.5+ billion tests annually, has achieved 110% average year-on-year growth over the last two years, and counts Microsoft, OpenAI, NVIDIA, and Vimeo among its enterprise customers.
Recognized in the 2025 Gartner® Magic Quadrant™ for AI-Augmented Software Testing Tools and The Forrester Wave™: Autonomous Testing Platforms 2025, TestMu AI has raised nearly $100 million in funding, including a $38 million Series C in 2024. Its annual TestMu Conference remains one of the industry’s most influential forums for AI-driven quality engineering.