Podcast Summary

Lenny's Podcast: Product | Career | Growth

Episode: The Non-Technical PM’s Guide to Building with Cursor
Guest: Zevi Arnowitz (Meta)
Host: Lenny Rachitsky
Date: January 18, 2026

Overview

This episode is a deep dive into how non-technical product managers (PMs) can leverage AI tools—specifically Cursor and Claude code—to build, ship, and iterate on real software products without writing or deeply understanding code. Guest Zevi Arnowitz, a PM at Meta (previously at Wix), shares his actionable, step-by-step workflow for "vibe coding," demonstrating how AI can transform every part of a product manager’s toolkit and unlock new possibilities for non-technical builders.

Key Discussion Points and Insights

1. The Non-Technical Builder Mindset

• Zevi's Background: Zevi has "zero technical background" ([06:05]), coming from music and psychology, not computer science.
  • "I have zero technical background. Did music in high school…" ([06:05] Zevi)
• AI as a Superpower: AI levels the playing field for non-engineers.
  • "It basically felt like someone came up to me and said you have superpowers now." ([06:05] Zevi)
• Democratization of Building: With AI, anyone curious and willing to learn can become a builder.
  • "I think everyone's going to become a builder. Titles are going to collapse and responsibilities are going to collapse." ([22:22] Zevi)

2. Zevi’s Evolving Tech Stack and Workflow

• Progression of Tools:
  • Started with project-based GPT tools (ChatGPT, Claude projects) for compartmentalized, chat-based learning and building.
  • Graduated to Bolt and Lovable (no-code AI app builders).
  • Ultimately moved to Cursor with Claude code for advanced control and flexibility.
• AI as CTO: Early on, Zevi used a GPT “project” with a custom prompt as his acting CTO.
  • "I created a CTO with the custom prompt of it being the complete technical owner of the project" ([07:56] Zevi)
• Key Advice: Start with simple UI-based chat tools for exposure therapy and work your way up to more code-heavy environments like Cursor ([13:18] Zevi).

3. Slash Commands: Structured, Repeatable AI Prompts

• Slash Commands: Prompts saved as reusable commands in Cursor, acting as workflow building blocks for common tasks ([14:46] Zevi).

  • Examples:
    • /create issue: Quickly captures bugs/features as Linear tickets.
    • /exploration phase: Prompts Claude to deeply understand a new issue.
    • /create plan: Outlines implementation as a markdown plan.
    • /execute: Tells an LLM (like Composer/Gemini) to build the spec’d feature.
    • /review and /peer review: Facilitates multi-model code reviews.
    • /learning opportunity: Explains complex code/concepts for learning.
  • Downloadable Prompts: All Zevi’s slash commands are available for listeners to use ([19:30] Zevi).

  Memorable quote:

  "What you describe here...is essentially what you've figured out is a way as a non...a person that has no idea how to write any code, how to build a product in cursor as a product manager, using this series of slash commands that you've concocted that you're going to be sharing with listeners."
  — Lenny ([19:00])

4. Workflow Demo: Shipping "Fill in the Blank" Questions in Studymate

• Studymate: Zevi showcases his AI-built student quiz app.

  • He walks through creating, planning, and shipping a new feature—30% of questions as "fill in the blank" with drag-and-drop UI—live on air.
  • Process:
    1. /create issue: Dictates the feature in natural language (voice)
    2. Claude auto-generates a Linear ticket and clarifies requirements ([22:43] Zevi).
    3. /exploration phase: Claude investigates current code, asks smart clarifying questions ([24:44] Zevi).
    4. /create plan: Converts understanding into a multi-step markdown plan ([28:17] Zevi).
    5. /execute: Sends plan to Composer/Gemini/Claude code to implement ([31:20] Zevi).
    6. Manual QA, then runs /review and /peer review for multi-model code review.
    7. Updates documentation for future agents ([48:00] Zevi).

• Multi-Model Review:
  Zevi involves Claude, Codex (OpenAI's code model), and sometimes Gemini for review and bug catching—with each model playing a unique "persona":

  • "Claude...would be the perfect cto... Codex...the best coder within the company who comes to the office with a hoodie and sandals and sits in a dark room... Gemini is like a crazy scientist..." ([39:08] Zevi)

5. Scaling, Code Review, and Quality

• Multi-Agent Peer Review:
  Models review each other's output, and Zevi plays them against each other:
  • *"It's really cool because...I can really tell you how each one of these would be as real humans...I'll have each of them review the code. Then...peer review, which...has them fight it out basically" ([39:08], [43:45] Zevi).
• Postmortems and Continuous Improvement:
  • When something fails, Zevi asks the model to identify the cause, then updates documentation or the prompt so it won’t happen again ([47:38] Zevi).
    • "...ask it, what in your system prompt or tooling made you make this mistake? And Cloud will...think of what made it create that mistake...let's update your tooling and documentation so that this mistake never occurs again." ([47:38] Zevi)

6. Learning, Growth, and the Future of Work

• Slash /learning opportunity: Invokes teaching moments mid-workflow.
• AI as a Learning Accelerant:
  • “Especially for more junior PMs, it allows you to play at such a higher level than you would normally.” ([54:25] Zevi)
• PMs Should Own Their Output:
  • “If you put anything out there...and you say, ‘sorry, that was built by AI,’ that’s your mistake.” ([54:25] Zevi)
• Advice for Large Companies:
  • Prepare the codebase to be “AI native” with clear docs and structure for agents.
  • PMs should experiment with contained UI projects, create PRs, and let engineers review before merging ([51:27] Zevi).

7. Meta Interview Prep—AI as Career Accelerator

• Used AI as Interview Coach:
  • Built a Claude project with PM interview frameworks, simulated interviews, generated feedback, and even constructed a quiz game to shore up weaknesses ([58:37] Zevi).
  • “Every time I'm faced with a new challenge or problem, I think AI first how to solve it.” ([58:37] Zevi)

8. The “AI Slop” Problem

• How to Reduce Low-Quality Output:
  • Be intentional with context, style, and guidance.
  • Update prompts as you learn.
  • Cursor’s /dslop command can refine messy code/output ([57:12] Zevi).

9. Zevi’s Big Advice for Aspiring Non-Technical Builders

• “If people walk away and open their computer and start building, you’ve succeeded.” ([06:05] Zevi)
• "It's the best time to be a junior. Contrary to what a lot of people are saying...when else in history could you get out of school and just build a startup on your own?" ([49:28], [66:25] Zevi)
• “If you're a kind person and a good communicator, you have such an unfair advantage and you can give more value to companies than most people who have 20 years of experience.” ([66:25] Zevi)

Notable Quotes & Memorable Moments

• AI Democratizes Building:

  "If you're non technical like me, code is terrifying. But AI just makes so much possible."
  — Zevi ([22:22])

• On PMs and AI:

  "It's not that you will be replaced by AI, you'll be replaced by someone who's better at using AI than you."
  — Lenny ([00:52], [62:33])

• On Learning Over Doing:

  "No one expects you to know all the answers and no one expects you to be good...be the best learner you can be."
  — Zevi ([63:26])

• On AI Feedback:

  "I want you to make me as ready as possible for these interviews. So give me feedback...And the other thing that I did was...I would ask Claude to play the candidate, and then it would just give me a really good answer."
  — Zevi ([61:21])

• On Documentation and Prompts:

  "Updating documentation and tooling is one of the biggest hacks for productivity."
  — Zevi ([45:54])

Timestamps of Important Segments

• Background—Zevi’s Zero-Tech Origin Story ([05:06]–[07:27])
• AI as CTO & Compartmentalized Projects ([07:56]–[11:40])
• Advancing from No-Code Tools to Cursor and Claude Code ([13:18]–[14:42])
• Detailed Workflow & Slash Commands ([14:46]–[19:54])
• Live Feature Build in Studymate ([22:30]–[38:56])
• Multi-Agent Code Review and “Peer Reviews” ([38:56]–[45:31])
• Postmortems and Documentation as Productivity Hacks ([45:54]–[48:00])
• Discussion: Impact on PM Craft, Reducing AI Slop ([54:25]–[57:12])
• Meta Interview Prep with AI ([58:12]–[62:20])
• Advice for Junior PMs and The Future ([66:25]–[67:43])
• Lightning Round: Books, Shows, Life Mottos, Entrepreneurial Stories ([67:54]–[73:30])

Key Takeaways for Non-Technical PMs & Builders

• Start Gradual, Start Simple: Begin with chat-based AI tools for ideation and learning; progress to more powerful tools like Cursor as you build confidence.
• Document Everything & Build Reusable Prompts: Systematize your workflow with slash commands to reduce repetition and human error.
• Embrace AI as Colleague(s), Not a Replacement: Use multiple models for “peer review” and play to each model’s strengths; imagine them as complementary team members.
• Update Your Processes Relentlessly: Do postmortems, refine prompts/tooling, and document lessons to drive compounding improvement.
• Leverage AI for Learning and Accelerated Career Growth: Use “learning opportunity” commands, run mock interviews, and simulate real-world tasks to climb faster.
• Don’t Fear the Future—Shape It: Zevi’s story proves that hard work, curiosity, and openness to learning outcompete deep technical knowledge in the era of advanced AI tools.

Practical Resources

• Downloadable Slash Commands/Prompts: Available via the episode show notes, ready to import into Cursor.
• Studymate: Zevi’s demo app for students—try it and send feedback.
• Connect with Zevi: Reach out on LinkedIn or X for help or collaboration.

Inspiring Closing Thought

“If people walk away and open their computer and start building, you’ve succeeded.”
— Zevi Arnowitz ([06:05])