The non-technical PM’s guide to building with Cursor

Zevi Arnovitz, a PM at Meta and former PM at Wix with no formal technical background, explains how he builds real products using Cursor and Claude Code. The conversation walks through his command-driven workflow for going from idea to Linear ticket, planning, building, peer review with multiple models, and documentation updates, plus his views on AI and the PM role.

01Key learnings

• Start slowly with a ChatGPT project acting as a skeptical 'CTO' before moving to coding tools, gradually easing into code to reduce the fear of looking at it.
• Capture ideas quickly as Linear issues using reusable slash commands, so you can pick them up later with context already gathered.
• Separate the workflow into explore, plan, execute, review, and document phases, and have the AI ask clarifying questions before writing any code.
• Use multiple models to peer review each other's code, since different models catch different issues; Zevi has them 'fight it out' until no issues remain.
• After mistakes, ask the model what in its prompts or tooling caused them, then update documentation so the same error doesn't recur.
• Use a learning-opportunity command to get 80/20 explanations of architecture when something is unclear, treating the AI as a patient mentor.
• Avoid shipping heavy database or migration changes as a non-engineer; contained UI work with a PR handed to developers is a realistic starting point.
• Being a 10x learner matters more than being the smartest person in the room; AI can help you get reps and play at a higher level earlier in your career.

Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.