#016 • 15 Apr 2025

Contents of this Newsletter

🧑🏻‍💻

AI Skills: Your Competitive Edge

⏰ : 5 min

📓

Blog Highlights

⏰ : 8 min

🪆

Team Action Figures

⏰ : 3 min

🗞️

Weekly AI News

⏰ : 4 min

MAIN STORY

Is AI The New Job Requirement?

Last week, a leaked internal memo from Shopify’s CEO Tobi Lütke made waves across tech. The message was that AI isn’t a side project anymore, it’s the job. AI proficiency is now tied to performance reviews, and teams must justify hiring a human over using AI. It’s bold but not surprising to us.

At Quash, we’re all in on this shift. This isn’t about chasing trends. It’s about acknowledging a new baseline: If you're not learning how to work with AI, you're actively falling behind.

AI Being the New Normal

Shopify isn’t alone. Artisan’s “Stop Hiring Humans” billboard campaign might’ve been designed to provoke, but the underlying question is real: what roles do we still need humans for? And more importantly, how do we equip them to thrive?

The numbers back it up. In 2024 alone, over 151,484 tech workers were laid off. At the same time, AI tooling reached new levels of maturity, enabling leaner teams to do more with less. And just last month, ChatGPT outpaced TikTok and Instagram to become the world’s most downloaded app—with 46 million installs in March, up 28% from February. The writing’s been on the wall and Shopify just had the guts to say it out loud.

At Quash, AI Is Already the Default

We’ve built Quash around the belief that AI shouldn’t just support your work, it should reshape it. That’s true in our product, and it’s true in how we operate internally.

Across our engineering, growth, and product teams, AI is part of the workflow—not an afterthought. Right now, we’re a 10x team across all departments and this is promising and possible—thanks to AI. Our developers rely on AI not just for debugging or suggestions but even for extensive code refactoring. This is something that would have taken at least a month, but is being done in just a week. We have a growth team consisting of just Prakhar, Donna and Mahak—but they overlook sales, campaigns and advertising, content, SEO, CRM, email and social media marketing, events, partnerships etc. All these tasks would at minimum demand a team of specialists but with AI it isn’t just doable, it's being done remarkably well.

And yes, we still share AI news, tools, and breakthroughs every day through our internal mailer. This isn’t one of those mailers that people ignore, we have full fledged discussions regarding these updates. It’s one of the ways we stay updated.

What the Future of Work Looks Like (From Where We’re Standing)

A recent study by Anthropic looked into how students are already using AI in their academic workflows and the results are telling. STEM students are leading the charge, with Computer Science students making up just 5.4% of U.S. degrees but accounting for nearly 40% of conversations with Claude.

That’s not curriculum-driven. That’s a culture being born. These students aren’t being forced to adopt AI. They’re reaching for it because it works. Whether it’s to vibe-check their code, debug faster, or ideate more freely, AI is quickly becoming second nature. In many ways, the next wave of developers won’t need to be told to learn AI—it’ll be the default.

AI is shifting what it means to be “skilled.” It’s no longer about knowing all the answers—it’s about knowing how to ask the right questions, structure the right prompts, and supervise with clarity. We’re seeing it in QA already. The best testers aren’t the ones manually clicking through flows—they’re the ones thinking critically, guiding AI agents, spotting gaps, and adapting quickly. That’s the future: not AI vs. human, but AI-literate humans doing more meaningful work.

And yes—some roles will vanish. Others will evolve. But the idea that we should hold off on this shift to protect outdated processes? That’s more of a denial than innovation.

Whether you’re leading a team or starting your first job, the AI requirement is a baseline expectation. The future of work won’t be built by those who resist AI. It’ll be built by those who figure out how to work with it—intentionally, intelligently, and with a bias for speed.

We’re already on that path. Let’s see who joins us.

Check out our Recent Blogs

Inspect Elements on Mobile Devices: A Step-by-Step Guide for Developers

If you've ever thought, "How do I inspect element on an App on Mobile Device...

Read article

Getting Started with BDD Testing: Benefits and Best Practices

Explore Behaviour Driven Development (BDD) Testing and its role in enhancing...

Read article

CReative

Imagining the Perfect Engineering Team

What happens when developers, QA engineers, and engineering managers finally feel at peace with their testing stack?

Introducing our limited-edition Quash Action Figures: a playful tribute to the people who build, break, and lead modern engineering teams, now stress-free, bug-free, and sipping chai.

Each figure comes with signature accessories like sticker-covered laptops, dashboards showing “0 blockers,” wired headphones and even post-its that proudly say “Quashed it!”

The idea for this came to us after a creator named Humberto went viral on X from early April, where he reimagined a series of politicians tied to Portugal’s upcoming elections as action figures—complete with satirical accessories and sharp commentary. We saw it, loved it, and thought: what if engineering teams had their own collectibles?

It’s a fun twist, but a real reflection of what we’re building toward at Quash — a world where testing isn’t a bottleneck, it’s just handled.

news

This Week in AI

OpenAI Launches GPT‑4.1 Family: Smarter, Cheaper, and Context-Savvy

OpenAI phases out GPT-4.5 from API access while introducing new GPT-4.1 models— GPT‑4.1, GPT‑4.1 mini, and nano in the API featuring major upgrades in coding (SWE-bench: 54.6%), instruction following (+10.5% over GPT-4o), and long-context reasoning (up to 1M tokens). These models outperform GPT‑4o across the board, with faster performance and improved real-world usability, now available at a lower cost.

Google Launches ADK: Framework for Building Powerful AI Agents

Google has unveiled ADK (Agent Development Kit) — a Python framework to build, test, and deploy multi-agent systems using Gemini and third-party models like Claude and LLaMA. With features like recursive tool use, model fallback, and structured workflows, ADK brings production-ready, tool-using agents to developers—fully integrated with Google’s Vertex AI stack.

GitHub MCP Server Now Open Source

GitHub releases its MCP Server in public preview—rewritten in Go with help from Anthropic. Supports code scanning, tool customization, and a new get_me function for natural language queries like “show me my private repos.” Fully integrated with VS Code.

OpenAI Launches Evals API

The new Evals API lets developers programmatically define and run evaluations for model outputs. It’s designed to test how well your LLM meets custom style, accuracy, and performance criteria, which is key for teams refining or upgrading AI apps.

Firebase Studio Debuts for AI Devs

Google introduces Firebase Studio, a cloud-based agentic IDE powered by Gemini. It’s built to help devs ship production-ready AI apps faster—no stitching tools, just full-stack speed. It’s a complete overhaul of how AI apps are built—bringing app creation, deployment, and data integration into one unified interface.

ElevenLabs Launches MCP Server

Now you can access ElevenLabs’ full AI audio suite—TTS, STT, voice design, and outbound calls—via text prompts in Claude, Cursor, and more. ElevenLabs is embracing the Model Context Protocol (MCP) with a new server that makes audio workflows instantly accessible to LLMs.

Smarter Memory in ChatGPT

ChatGPT can now remember everything you’ve shared in past chats, enabling more personalized, context-aware responses. Whether you’re writing, learning, or just exploring, it tailors its support based on your preferences and history.

Anthropic Finds Gaps in Chain-of-Thought Reasoning

Only 25–39% of reasoning models mention all their key internal thoughts in Chain-of-Thought (CoT) explanations, raising safety red flags. This limits how much we can trust CoT for alignment and transparency.

AI in Performance Reviews + Engineering Heroes