#028 • 8 July 2025

Contents of this Newsletter

⭐️

10,000 Test Cases and Counting

⏰ : 6 min

📓

Blog Picks

⏰ : 11 min

🧠

Soham Parekh for Testing

⏰ : 3 min

🗞️

Weekly AI News

⏰ : 6 min

MAIN STORY

10,000 Test Cases Later

Every once in a while, something small turns out to be bigger than it seemed. Not because it made noise, but because it worked.

It wasn’t that long ago that we were stitching up half-written PRDs and early Figma flows, trying to see if an agent could even click a button correctly. We weren’t sure this approach would work. We just knew QA needed a different kind of foundation and that is one built on context.

That word ‘context’ kept showing up. In product reviews, design critiques and test failures. It’s also what kept showing up in this newsletter, week after week, as we started shaping something we hoped could live up to it.

Eventually, we shared a small piece of that work: a test case generator. There wasn’t a big launch or anything. Just a link in the newsletter. A tool for teams to try out, for free.

And while that tool started making its way around, across our customers, teams were using Quash with their own PRDs, Figma flows, and integrated test pipelines. They got relevant and execution ready test cases. They were test cases that worked and ran on actual devices.

And last week, we crossed the generation of 10,000 test cases.

Ayush Shrivastava

@ironmark1993

wasn’t that long ago we were stitching together figma dumps and half-written PRDs wondering if an agent could even click a button.
today it crossed 11,276 test cases.
no fanfare. no launch(but i think we should).
just this quiet, weird feeling that we might’ve bent something

7:14 PM • Jul 2, 2025

Retweets

Likes

Read 1 replies

That shift to context-aware, intelligent QA is now more than a feature. It marks a turning point for us. It’s not just about better capabilities; it’s about making them accessible. Quash has long been enterprise-first. Now, with the right foundations in place, we’re officially moving toward General Availability.

It’s about access and we’re opening Quash up to everyone who builds software: solo developers, startups, small QA teams, and fast-moving product orgs.

From Bug Reporting to Test Intelligence

Quash didn’t start with a grand roadmap. It started as a side project—a bug reporting SDK that manual testers could drop into their apps to capture bugs with context. It was open source, and more than 2,000 developers picked it up. It found its way into open-source projects, indie teams and internal QA pipelines. It worked because it made feedback easier, faster, and more structured.

Then came our push into AI-powered, CI-integrated testing. We built it for speed. Teams could generate tests from code, PRDs, and designs, then plug them straight into GitHub, GitLab, and other CI providers. It scaled well. Enterprises used it to validate every release across platforms. But we realized something was still missing.

Speed wasn’t enough. What QA really needed was context.

That’s where the test case generator comes in. It doesn’t rely on vague prompts or assumptions. It understands what’s in your PRD. It reads the copy inside your Figma. It knows that a feature flow doesn’t live in isolation, but it’s shaped by product decisions, design nuances, and release constraints. That’s the shift that matters: from automating effort to generating insight.

Most tools in QA ask you to move faster. We wanted to help teams start smarter.

Moving Forward

Crossing 10,000 test cases wasn’t just a milestone. It was a mirror.

It showed us how teams were using Quash, to shift the role of QA itself. They weren’t waiting until the end of the release cycle. They were starting at the beginning—at the planning doc, the design file, the repo. They were putting test coverage closer to product thinking.

That’s not something we predicted. But it’s something we’re ready for.

Because this spike in usage was a signal. A quiet but clear one that the idea of context-aware QA isn’t niche. It resonates, works and it’s needed.

We still have a lot to build, but we’re at the point where we can start opening things up.

The generator isn’t live for everyone yet. But we’re getting there. The waitlist is now open, and we’re inviting the kinds of teams who want QA to think earlier, act smarter, and do more than just catch bugs.

If that sounds like you, join the waitlist. We’d love to have you onboard early.

And to the team that brought this to life, from bug reports to runnable test cases, across real devices and real-world constraints, we couldn’t be prouder of how far this has come.

Check out our Recent Blogs

Team

We’re Building Soham Parekh for Testing

Last week, the story of Soham Parekh hit the tech world like a glitch in the Matrix. A single engineer holding 34 full-time jobs at once, across dozens of startups, quietly collecting paychecks while shipping just enough. At first glance, it feels like fraud. But dig deeper and you’ll find something messier: a guy who understood the inefficiencies of modern work better than most of the companies hiring him.

Soham didn’t scale himself with AI, or a team of ghostwriters, or time zone tricks. He just knew how to play the system—show up for interviews, dazzle with pair programming, then do the absolute minimum to avoid suspicion. He didn’t automate the work, he optimized his presence. And weirdly enough, that’s exactly what good QA should look like: show up where it matters, stay out of the way otherwise.

At Quash, we’re not here to condone moonlighting. But we do admire the principle behind Soham’s hustle: invisible efficiency. Our CEO put it bluntly—“We’re building Soham Parekh for testing.” Not a person, but a presence. An agent that blends into your team, reads your PRDs and designs, flags the important stuff, and then disappears. No standups, no ten different JIRA tickets saying the same thing. Just context-aware QA that knows when to act.

Soham scaled himself across 34 startups. We’re scaling QA across all of yours.

news

This Week in AI

Grok 4 Launches July 9 with Multimodal AI Upgrades

xAI will unveil Grok 4, its latest AI model, via livestream on July 9 at 8 PM PT. Elon Musk confirmed the launch after previously hinting at a post–July 4 release. While specifics remain under wraps, Grok 4 is expected to bring major upgrades in reasoning, coding, and multimodal features like vision and image generation. Musk also teased new capabilities, including meme comprehension—"Grok understands memes. Think about that."

Apple Considers Anthropic or OpenAI to Power Siri, Bloomberg Reports
In a shift from its in-house AI efforts, Apple is reportedly in early talks to use Anthropic or OpenAI’s models to power a future version of Siri. According to Bloomberg, Apple has asked both companies to train models on its cloud for internal testing. This comes after delays in Siri’s AI upgrades and leadership changes within Apple’s AI team. While no final decision has been made, the move signals Apple’s openness to external LLMs, following its recent announcement to support both its own and OpenAI’s tools in developer software.

Cursor Agents Now Available on Web and Mobile

Cursor has expanded its powerful coding assistant to web and mobile. Just like in the IDE, Cursor Agents can now write code, fix bugs, and handle complex dev tasks from any browser or device. You can launch tasks, review PRs, and collaborate with teammates on the go. Bonus: it works as a PWA for a native-like experience on iOS and Android. Start exploring at cursor.com/agents.

All Newsletter Issues | Security | Contact Us | Blogs | Test Case Generator

The Feature We Didn’t Launch, But Everyone Used 😳