scripod.com

The rise of the professional vibe coder (a new AI-era job) | Lazar Jovanovic (Professional Vibe Coder)

This episode features Lazar Jovanovic, a professional 'vibe coder' at Lovable—someone who builds production-grade internal and customer-facing tools using AI, despite no formal coding background. He shares how he navigates the evolving landscape of AI-augmented development not as a programmer, but as a strategic thinker, designer, and product-minded builder.
Lazar emphasizes that non-technical backgrounds can be an advantage in AI-driven development—freeing builders from legacy constraints and fostering imaginative, user-centered solutions. He advocates spending 80% of time on planning, context-setting, and collaborative ideation in chat mode—not prompting or coding—and introduces his 'genie and three wishes' mental model to stress precision in intent communication. His workflow includes parallel prototyping, dynamic PRD-based context windows, and structured Markdown files (e.g., masterplan.md) to align AI agents across complex projects. He details a 4x4 debugging framework, highlights the growing importance of taste, judgment, and emotional intelligence over syntax mastery, and argues that 'good enough' is obsolete—only world-class design stands out. As engineering roles converge with product and design, success increasingly hinges on clarity, quality, and building in public—not technical pedigree.
00:00
00:00
Vibe coding merges engineering, design, and product management into one AI-augmented role
04:56
04:56
Vibe coding is a dream job using Lovable to push projects to production across internal and external tasks
09:26
09:26
The key is to have a delusional belief that everything is possible
12:24
12:24
Spend 80% of time on planning and chatting in chat mode and 20% on execution
14:42
14:42
While the token limitation can't be controlled, humans can improve by being specific and providing context
17:44
17:44
Clarity in AI prompting involves understanding taste, quality levels, and what's magical
27:18
27:18
Parallel prototyping helps clarify ideas, avoid being locked into the first design, and save money by getting things right from the start
35:55
35:55
Focusing 100% on skills like good judgment, clarity, taste, and using good fonts, as these require better decision-making
42:18
42:18
The ceiling of AI performance is what it sees before acting
44:43
44:43
Masterplan.md provides a high-level overview of the app's intent, who it's for, and references other PRDs
50:58
50:58
Lovable enables non-engineers to build internal tools with AI
56:57
56:57
AI is an amplifier, and 'good enough' is no longer sufficient as everyone can produce it with AI
1:00:55
1:00:55
Elite engineers will be needed more than ever for codebase maintenance, infrastructure building, and handling issues like internet outages
1:13:41
1:13:41
Codex can solve the most difficult bugs when given code and console output
1:14:27
1:14:27
Ask the AI agent how to prompt it better, then record lessons in rules.md to eliminate human error and let the AI learn from them
1:18:31
1:18:31
AI's feedback refines judgment and planning clarity more than technical skill alone
1:19:08
1:19:08
AI is causing rapid job role evolution, rendering many previous workarounds and teachings obsolete
1:26:05
1:26:05
AI will replace translators and most journalists, but not comedians—it can't write good jokes
1:28:30
1:28:30
Past experiences—blue-collar jobs, forestry engineering, waiting tables, and startup roles—all prepared me for my dream job, like 'Slumdog Millionaire'
1:35:44
1:35:44
Fear will turn to excitement if you just try building something
1:37:07
1:37:07
In the AI era, tech stack matters less, and the end-user wants a great experience