The Future of Testing Isn’t Recorded — It’s Reasoned: My Chat with Marcel from Wopee.io

pizzacheese xl
May 12
3 min read

🚀 The Future of Testing Isn’t Recorded — It’s Reasoned: My Chat with Marcel from Wopee.io

By Mike Verinder, Founder of Modern Software

If you've ever wondered where autonomous testing is really headed — beyond the demos, buzzwords, and vaporware — you need to hear what Marcel and his team at Wopee.io are building.

I had the chance to sit down with Marcel, and what followed was an unfiltered, technical, and often philosophical conversation about the state of test automation, AI agents, and how much PowerPoint it really takes to launch a startup.

You can watch the full conversation on YouTube, or keep reading for the highlights.

🧠 From PowerPoint to Product

Marcel’s journey started where most great test tools don’t — with a question, not a solution. Years ago, a customer told him, “We’ve got automation figured out… what’s next?” That question sparked years of exploration, prototyping, and eventually: Wopee.io — a product built from a PowerPoint deck that customers were literally asking to buy before it even existed.

What sets Wopee apart is its goal: not to replace testers, but to boost productivity by combining AI agents with Playwright-based automation. It’s automation that reasons — not just records.

🕵️‍♂️ How It Works: Agents, Not Scripts

Wopee’s agent-based model breaks tasks into intelligent steps — like logging in, navigating, asserting conditions — and uses LLMs to orchestrate tools in real-time. You can prompt it, steer it, or just give it a URL and let it work.

Under the hood:

Crawling → Context → User Stories → Test Cases → Playwright Code.
Then? You own it. Export the code, plug it into your repo, or use it as-is in CI.

It's not magic. It’s test scaffolding on-demand, with built-in flexibility and explainability.

🧪 Visual Testing: The Underrated War Zone

Marcel didn’t hold back on visual testing — calling out most tools for being either too shallow or too complex. Wopee is intentionally on the “light” side, aimed at simple screen comparisons, fast feedback, and practical test validation — not pixel paranoia.

But here’s where it gets interesting: they’re experimenting with AI-driven visual assertions — asking things like “Is the login button visible?” rather than just diffing screenshots. That’s next-level validation.

🤖 On Agents, MCP, and Why They’re Waiting

We talked Model Context Protocol (MCP), which I love — but Marcel’s take was refreshingly honest: it’s too early. Instead of blindly integrating the latest AI spec, Wopee is laser-focused on what customers need now — and building toward a truly autonomous testing assistant, not just a wrapper around LLMs.

His advice? Don’t get distracted by hype. Let technology mature. Build what works first.

💡 Who Wopee Is For

Wopee.io isn’t trying to be all things to all testers. It’s for teams who:

Want to bootstrap automation from a URL.
Need a jumpstart for Playwright-based testing.
Are tired of brittle scripts and want AI-assisted scaffolding.

And while it’s early days for mobile and API coverage, those areas are on the roadmap — but only when they can be done right.

📌 Final Thoughts

What I appreciated most about this chat was Marcel’s honesty. No product fluff. Just clear thinking, thoughtful roadmap decisions, and a vision for a future where AI assists testing without taking over.

If you’re serious about the future of test automation — and especially if you’ve been burned by tools that promised “autonomous” and delivered... less — you should give Wopee.io a look.

👉 Watch the full interview here: https://youtu.be/q5NccmMAXJ0

👉 Explore the product: https://wopee.io