#049 • 02 December 2025

Contents of this Newsletter

📈

AI Testing Hits Inflection Point

⏰ : 5 min

🙏🏼

Join Us For An Event This Week!

⏰ : 3 min

✅

Quash Tests Your Apps

⏰ : 3 min

🔔

What We Shipped Last Week

⏰ : 2 min

📓

Blog Picks

⏰ : 10 min

🗞️

Weekly AI News

⏰ : 9 min

MAIN STORY

AI Testing Is Hitting Its First Real Inflection Point. Here’s What’s Actually Changing.

For years, the idea of “AI in testing” has floated around in discussions, conference slides and vendor marketing pages, but if you worked inside an engineering team, especially mobile, anytime between 2018 and 2022, you probably remember that most of these claims did not translate to real productivity gains. Teams were still fighting flaky tests, maintaining selectors and juggling device coverage the same way they had been for a decade. The tools talked about intelligence, yet the workflows still felt like automation with a few helpful shortcuts. This year, however, the conversation inside mobile teams and QA groups feels different, and the shift is not being driven by hype but by pressure and maturity across the ecosystem.

The Landscape Has Quietly Split Into Four Camps

If you map the current tool ecosystem, it becomes clear that the broad category of “AI testing” is now four very different approaches. The first camp contains the true agentic tools, where products like SpurTest and Finalrun attempt to behave like a QA engineer by exploring flows from high-level goals rather than step-by-step scripts. The second camp consists of large device cloud providers such as LambdaTest and BrowserStack that have added AI as layers on top of their existing infrastructure, usually in the form of self-healing, flaky-test detection, failure insights, or visual analysis. The third camp holds the modern record-and-replay platforms like Mabl, Autify and QA Wolf, which continue to evolve low-code testing with AI-generated tests, improved locators, and DevOps integration. The final camp includes traditional frameworks like Cypress, Maestro, Espresso and Appium that now offer AI-assisted features through plug-ins or GitHub integrations, yet remain fundamentally code-first in how teams operate them.

Each camp solves a different part of the testing burden, but the important observation is that most tools still rely on AI as a helper sitting inside an existing workflow and only a very small set are trying to build agents that can operate across devices and complex mobile interfaces. Most of the activity in this space sits in those helper categories, and only a limited group of products are building true agentic systems that operate from high-level goals on real devices. Quash belongs in this camp, focused on mobile use-agents that perform actions the way a tester would rather than through predefined scripts. This is the first signal that the industry is entering a new phase.

Why This Moment Is Different

The real story is not that AI suddenly became capable of replacing testers. It is that several unrelated trends have converged to make agentic testing feel inevitable. Models have improved in reasoning and stability. Real device clouds have become cheaper and more accessible. Mobile teams are releasing faster than ever, with multiple industry reports in 2024 and 2025 showing that a majority of consumer apps ship at least monthly and many aim for weekly or biweekly cycles. Recent surveys show that agentic AI has moved beyond prototypes, with a McKinsey report noting that almost a quarter of organisations say they are already scaling an agentic AI system. The budget side is shifting as well, with a NASSCOM study showing that nearly 90 percent of enterprises plan to allocate specific budgets for testing or building AI agents and more than 60 percent have moved into pilot or production stages. This acceleration in workload without a corresponding increase in QA headcount is what pushed teams to look beyond traditional automation.

In other words, AI testing did not mature simply because AI became smarter. It matured because engineering teams ran out of human capacity to keep up with rapid release cycles, especially on mobile where set-up, gesture handling, network variability, and OS-level behaviors make scripted automation brittle and slow to maintain.

What Teams Should Look For and Where the Real Opportunity Sits

As more teams evaluate AI-driven testing, the most helpful criteria are surprisingly grounded. A solid platform should reliably handle authentication, adapt when the UI changes and run consistently on real devices rather than relying only on emulators. It should offer clear, actionable feedback when something breaks, which means providing useful failure context, automatic hints on likely root causes, flakiness signals and a way to replay the agent’s behaviour. Modern systems should also lighten test creation and upkeep through natural-language instructions, dependable self-healing, shared assets that reduce duplication and a clean integration path into CI pipelines and device clouds. These are the capabilities that separate tools that only work well in a controlled demo from those that genuinely reduce regression load across rapid release cycles.

The opportunity ahead is clear. Testing is gradually moving from scripts toward agents, but only the tools built with traceability, explainability and a strong understanding of mobile complexity will sustain long-term adoption. Teams need systems that help them test faster, understand failures more clearly and keep pace with the release cycles they are now expected to meet. If this moment feels different, it is because the pressure is real and the technology has finally reached a place where the industry can respond to it with something more substantial than a buzzword.

We’ve been building around this shift for a while, and as of today our GA version is live for anyone who wants to see what an agent built specifically for mobile testing looks like in practice. We’re also giving away free credits so teams can try it without any setup friction. If you’re curious, you can check it out here: https://quashbugs.com/

TEAM

Is the AI Wave in Testing Just a Hype?

We’re kicking off December with a session that tackles a question everyone in QA is quietly asking: Is the AI wave in testing actually real, or just another hype cycle? We’ve lined up speaker sessions featuring QA leaders from companies like Amazon, Groww and others who’ve been working with AI-driven testing inside high-scale engineering environments. The plan is to get past the surface-level chatter and talk about what’s genuinely shifting in mobile QA and how teams are adapting, what’s working (or not).

Along with the talks, we’re running a rapid-fire quiz, a short panel discussion, and then a live Quash showcase. After the demo, anyone attending can try the tool themselves in a hands-on segment with real devices set up on-site. It’s a chance to see how the agent behaves in actual app flows and understand where it fits in a modern QA workflow. If you’re curious about where testing is headed over the next year, this one should be worth tuning into. Link to register: https://luma.com/q2mztreo

Your Apps, We Test

Each week, we put our execution engine to work on real apps.

This week, we put our execution engine to work on BigBasket, India’s largest online grocery platform. Instead of a single prompt, we ran a full test suite to validate how different coupons behave at different cart values. The suite began by opening the BigBasket app, adding an avocado to the cart, and incrementing the quantity until the total reached between ₹199 and ₹299. Quash checked for the BHIM30 coupon, applied it, validated it, and then cleared the filter.

Next, the engine increased the cart value to the ₹299–₹399 range and verified that the NEWBBNOW coupon appeared and could be applied successfully. After that, it continued adding items until the cart crossed ₹399, checked for GPAYBB5, applied it, validated the application, and cleared the coupon filter once again.

To wrap up, the suite removed the applied coupon and closed the app, thereby completing an end-to-end validation of BigBasket’s coupon display and application logic across multiple thresholds.

Want us to test your app next? Drop us a mail at hello@quashbugs.com

What Did We Ship?

Test Data

v1.0.8 | December 2, 2025

This week, we shipped Test Data, which is a simple way to control exactly what inputs the agent should use before a run starts. Instead of editing prompts or tweaking flows every time you switch environments, you can now define your data upfront and let the agent handle the rest.

Whether it’s dev credentials, stage-only values, dummy phone numbers, or those random constants you usually copy from Notion, you can store them once and reuse them across tasks and suites. When a run begins, the agent automatically pulls the right values into the flow.

Check out our Recent Blogs

News

This Week In AI

Apple’s Major AI Leadership Shakeup
Apple quietly restructured its entire AI org this week. John Giannandrea is stepping back, and Amar Subramanya is taking over core AI responsibilities. It’s the clearest signal yet that Apple knows it’s behind in generative AI and is reorganising to close the gap fast.

Nvidia Invests $2B in Synopsys
Nvidia put $2 billion into Synopsys to deepen control over the chip-design ecosystem. This isn’t a normal investment, instead it ties AI hardware, model training, and EDA tools together, giving Nvidia even more leverage over the AI stack.

Australia’s National AI Plan Warns of Security Risks
Australia launched a national AI plan that essentially says: AI will accelerate national-security risks and governments need to act now. They’re setting up a government-level AI Safety Institute with dedicated funding, signalling that regulation and safety work is no longer academic; it's state-level priority.

All Newsletter Issues | Security | Contact Us | Blogs

Our GA Launch + Come See the Future of AI Testing on Dec 6 🔮

Product

Solutions

By role

By use-case

By industry

Resources

Comparison

Company

Product

Solutions

By role

By use-case

By industry

Resources

Comparison

Company

Get Quash

Product

Solutions

By role

By use-case

Resources

Comparison

Company

Get Quash