The Tech Was Easy. The Humans Were Hard.

Two product managers—one calling in from Beirut, the other splitting time between Paris and Dubai—sat down for a casual chat about AI, vibe coding, and the tools everyone is chasing. What emerged wasn’t just a laundry list of apps. It was a story about timing, hype, hard decisions, and the one universal truth: building software is simple compared to changing people.

The Rise and Plateau of Lovable

Lovable was never the best vibe coding tool. It didn’t matter.

It grew because it launched early, nailed its branding, struck the right partnerships, and moved at breakneck speed. The team shipped features like their lives depended on it, turned outages into marketing moments, and rode the vibe coding wave before anyone else was even looking.

That mix of timing, luck, and charisma made Lovable explode. But beneath the surface, the product wasn’t hard to copy. An LLM, a few prompts, some APIs—anyone could rebuild it in a weekend. Which is exactly why the space is now overcrowded with 100+ competitors.

In other words, the plateau was baked in from the start.

The Reality of AI in the UAE

Leaving Lovable wasn’t just a lifestyle decision; it was about impact. The move to the UAE government promised a chance to help build AI from the ground up in the region.

The reality? Mixed. Arabic NLP is strong—speech-to-text, text analysis, culturally specific use cases. But everything else lags far behind the US, Europe, and China. Image generation is weak. Productization is thin. The country is better at consuming AI than producing it.

And the people? That’s the real challenge. The money is there, the expertise isn’t. Even officials with “AI” in their title often don’t understand what a model actually does. Progress requires not just building tools, but shifting entire mindsets.

Vibe Coding in Practice

Meanwhile, the tool-hopping circus goes on. Cursor is the one constant—a reliable partner for bug-fixing and generation. N8N trumps Zapier because it’s cheaper, scalable, and self-hostable. Orchid, Makali, Rok—new platforms pop up daily, most offering just enough free credits to keep users dabbling.

Mobile remains a gap. Tools like Rok help transform web apps into mobile apps, but the quality isn’t there yet.

The point: vibe coding isn’t about fidelity. It’s about momentum. Developers, PMs, and designers all stitch together their workflows with whatever’s fastest that week.

Prompting as a Skillset

The real work isn’t clicking around in an interface—it’s structuring prompts and workflows.

• Start with a voice dump: talk through the whole idea, 20 minutes unfiltered.
• Triangulate: run prompts through OpenAI, Anthropic, Grok, let them compete, then combine the best.
• Break it down: don’t ask for A → B, ask for A1 → A5 → B. Smaller steps, better results.
• Structure everything: who you are, the context, the expected output.
• Build front-end first: mock data, clickable prototypes, polish the experience before back-end complexity.

This isn’t “prompt engineering” as a gimmick. It’s product sense, rephrased for machines.

Workflow Beyond Tools

And here’s where the human element slams back into focus.

Tools are cheap, replaceable, disposable. What moves organizations is showing—not telling. A clickable prototype will change more minds than a PRD ever will. A quick demo in a meeting can do more than weeks of persuasion.

Adoption is emotional. Developers fear being left behind, so they try Cursor. PMs crave influence, so they prompt their way into wireframes. Executives respond to stories of competitors moving faster, not diagrams of architecture.

The lesson is blunt: building the tech is the easy part. Convincing humans to change is where the real work begins.

The takeaway: Vibe coding tools, AI browsers, automated workflows—they’ll keep coming and going. What lasts is your ability to translate them into something people actually want to use, or better yet, fear missing out on.

Because at the end of the day, code is predictable. Humans aren’t.