The non-technical PM’s guide to building with Cursor | Zevi Arnovitz (Meta)
The non-technical PM’s guide to building with Cursor | Zevi Arnovitz (Meta)
The non-technical PM’s guide to building with Cursor | Zevi Arnovitz (Meta)
In an era where AI is reshaping how software is built, non-technical professionals are gaining unprecedented power to create real products. This episode dives into the practical strategies that enable product managers without coding backgrounds to design, develop, and ship applications using cutting-edge AI tools.
Zevi Arnovitz, a non-technical product manager at Meta, shares his end-to-end AI workflow using Cursor and multiple large language models to build production-grade apps. He demonstrates how to leverage different AI models for specific roles—Claude as a strategic planner, Gemini for UI design, and Codex for efficient coding—and how personifying them improves collaboration. A key innovation is his peer review system, where models critique each other’s code to catch bugs and improve quality. Zevi emphasizes structured workflows with slash commands, postmortems for continuous improvement, and AI-assisted learning, both for development and interview prep. His approach enables rapid prototyping and execution, turning ideas into shipped features in minutes rather than weeks. The discussion highlights how AI lowers barriers for non-engineers, transforms the PM role, and creates new opportunities for junior talent to thrive in tech by learning fast and building smarter.
13:26
13:26
Code is just words and files; projects can move across apps.
14:42
14:42
The 'create issue' slash command captures bugs or features directly in the codebase for later tracking in Linear
23:15
23:15
Claude reads the codebase and asks targeted questions to refine the development plan.
35:53
35:53
Cursor Composer can build full-featured apps in minutes compared to human engineers' weeks
38:33
38:33
Using Claude to review its own code, alongside tools like Codex and Cursor, improves code quality.
40:40
40:40
Claude is like a communicative and opinionated CTO, Codex a non-communicative problem-solving coder, and Gemini an artsy yet erratic designer.
43:37
43:37
Use '/learning opportunity' to understand unfamiliar coding concepts during review
45:40
45:40
When AI makes a mistake, ask it to reflect on the root cause and update your processes accordingly.
51:05
51:05
In a few years, everyone may become a builder as AI models improve.
53:43
53:43
Using AI intentionally allows junior PMs to operate at a higher level and gain valuable experience.
57:03
57:03
Use context and style guidance to improve AI results, similar to managing a junior colleague.
1:00:54
1:00:54
One may be replaced by those better at using AI in interviews.
1:04:00
1:04:00
He changed his mindset to become a '10X learner' and sought mentorship from experienced PMs
1:09:01
1:09:01
Nobody knows what the fuck they're doing
