#026 • 24 June 2025

Contents of this Newsletter

🏢

More Space, More Clarity

⏰ : 4 min

📓

Blog Picks

⏰ : 11 min

📝

iyO Sued io

⏰ : 2 min

🗞️

Weekly AI News

⏰ : 6 min

MAIN STORY

What Does Real Product Building Look Like?

This Week Was One Long Whiteboard

Most people only see what ships. The launch. The announcements. The polished story at the end. But what rarely gets shared is what it really takes to get there. This week at Quash was one of those weeks, where nothing shipped, but a lot moved.

The product landscape is moving fast. Every other week, a new AI agent, dev tool, or automation layer enters the arena. We reframed parts of the workflow, challenged how results surfaced, and even reconsidered what a “project” should feel like in the tool. These weren’t small changes. They were directional ones. Even seemingly small UX details like when to nudge someone for feedback became conversations about trust, defaults, and developer control. The team sat with decisions longer than usual because speed only works when precision comes with it.

A Change of Space

Alongside all this, we also moved. The old office in HSR Sector 2 was getting tight, too few rooms for calls, too little space to think. So we shifted to a new one in Sector 7. Bigger, brighter, and better suited for the pace we’re operating at. The move wasn’t just about expansion. It was about alignment. The space had to catch up with the team—not just in headcount, but in energy and intent.

Within a day, people were back in, setting up, scribbling on desks, resuming the same conversations where they left off. It felt like a physical mirror of what’s happening in the product, making room where it matters. Just like how we’re rethinking the structure of projects or how AI-driven results are presented, we’re also making space for sharper thinking in how we work.

Building With Skin in the Game

Quash started from a very real frustration: the mobile experience was moving fast, but testing never caught up. Most QA tools felt like an afterthought—slow, fragmented, and unfit for how quickly modern teams ship. And now, with AI accelerating code generation like never before, the gap between development and testing has only widened. That’s where we knew we had to step in. Our goal has always been simple: help teams release high-quality apps without QA slowing them down. Because quality isn’t just a technical checkbox; it’s the foundation of user trust, retention, and experience.

That mission also keeps us honest. Every product decision we make: from how onboarding works to how test results are surfaced, is a reflection of what we believe about how developers want to work. And what they’re willing to trust AI to do. Earlier this week, our CEO shared a thought that’s been echoing through the team ever since:

Ayush Shrivastava

@ironmark1993

the subscription economy trained an entire generation to think ownership is inefficient, but now that same generation cant afford houses, cars, or kids. we optimized for access over assets & accidentally created a permanent renter class that confuses convenience with wealth.

10:21 PM • Jun 20, 2025

Retweets

Likes

Read 2 replies

That idea: access without trust, cuts close to the work we’re doing. We’re not just trying to ship fast; we’re trying to ship right. That’s why we spend time on the details. And it’s why even the newest people on the team, including interns, are invited into product conversations early. They’ve asked sharp questions, surfaced blind spots, and helped us hold ourselves to a higher bar. Because real product building takes more than speed. It takes people who care enough to get it right.

Check out our Recent Blogs

social

The ‘io’ Name Isn’t Settled Yet

Last month, OpenAI and Jony Ive announced a new AI hardware venture called io, a screenless device that’s meant to quietly live on your desk and assist you with AI. But now, that name is facing legal trouble.

A startup called iyO has filed a lawsuit against OpenAI, Sam Altman, and Jony Ive, claiming that io is not just a similar name, it’s uncomfortably close to the product and vision IYO pitched them a few years ago. According to the filing, IYO had shared their early concept with both teams, hoping for a partnership. They passed.

Since then, iyO says someone linked to the new io team bought one of their devices, asked for deep product details, and later, Altman allegedly responded to a new funding inquiry with: “Thanks but I’m working on something competitive so will respectfully pass!”

The lawsuit covers trademark infringement and unfair competition. It doesn’t stop the project, but it does force a pause on branding. OpenAI has already pulled down the io landing page and related content, although the partnership still stands.

The product might still ship next year. But the name part will have to be sorted out.

news

This Week in AI

Anthropic Thinks LLMs Can Be Freaky
Anthropic dropped SHADE-Arena, a benchmark for testing if models can secretly complete harmful tasks while looking innocent. Top models like Claude Opus, Gemini 2.5 Pro, and O4-mini pulled off covert sabotage over 50% of the time, without raising suspicion. Human reviewers? Slower and less accurate. Only 5 of 17 test cases are public.

ElevenLabs Voice Agents Can Now Use Your Tools
ElevenLabs added support for MCP (Model Context Protocol), a standard that lets voice agents connect to apps like Gmail, Salesforce, and HubSpot. This means agents can now access data, perform tasks, and respond more intelligently—making them feel less like bots and more like helpful assistants that actually understand your tools.

OpenAI Finds and Fixes Misaligned Behavior in GPT-4o
Researchers discovered a “misaligned persona” inside GPT-4o—an internal feature triggered by bad fine-tuning data. Using sparse autoencoders, they isolated this pattern and reversed it by fine-tuning on just 120 good examples. The fix worked in just 30 steps and generalized across tasks.

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

Behind the Build: A Real Week at Quash + “io” Hits Legal Trouble ⚖️

Platform

Solutions

By role

By use-case

Resources

Comparison

Company

Platform

Solutions

By role

By use-case

Resources

Comparison

Company

Get Quash

Platform

Solutions

By role

By use-case

Resources

Comparison

Company

Get Quash