Listen to this episode on: Spotify | Apple Podcasts
Petra finally carved out time to try Claude Code—and immediately ran into the “this tool thinks everything is a code repo” reality. In this episode, Petra and Teresa unpack what it’s like to move from Claude-in-the-browser to Claude on your machine, where it can actually work with your files, folders, and workflows.
They dig into practical setup choices (like using /init, Claude MD rules, and “walled garden” access), why task management becomes the surprising on-ramp for testing Claude Code, and how to use it for content retrieval across your own archives (blog posts, book drafts, transcripts, notes). Teresa also shares how she’s turning Claude Code into a repeatable publishing engine—from podcast metadata generation to fact-checking and even building a Zettelkasten-style research system for rigorous thinking.
If you’re a non-engineer curious about Claude Code, or you’re trying to make it feel less like “throwing a stick into the woods,” this is a candid, tactical walkthrough of what helps, what breaks, and how to debug your way to workflows that actually stick.
What we cover
- What Claude Code is (and why it’s useful even if you’re not an engineer)
- The real difference between Claude in the browser vs. Claude Code on your machine
- Using /init and Claude MD files to shape behavior and context
- The “treat it like an intern” model: limiting access and building a walled garden
- Why task management is a great first use case (easy to validate + fast feedback loops)
- Exporting calendar windows vs. connecting Claude directly to your calendar
- Switching models for speed: when Haiku is “good enough” and when you need more
- When Claude Code spirals on web tasks (and how to debug it by asking: “What are you doing?”)
- Content retrieval as a killer workflow: “Where have I talked about this before?”
- Building reusable workflows: slash commands, sub-agents, hooks, and plugins
- Teresa’s publishing stack: titles, descriptions, show notes, chapters from transcripts
- Fact-checking workflows: validating claims against transcripts and research sources
- Using Claude for audience analytics and content prioritization (with caveats)
- A deep nerdy detour: Zettelkasten-style research and using Claude to improve rigor
Memorable moments / quotes (paraphrased)
- Petra’s metaphor: Claude Code is like a dog—sometimes it returns with the stick, sometimes it gets lost in the woods.
- The “intern” framing: don’t hand Claude the whole company on day one—scope access intentionally.
- Teresa’s output jump: more writing volume without (in her view) losing quality—because the workflow scaffolding got better.
Resources & Links:
- Follow Teresa Torres: https://ProductTalk.org
- Follow Petra Wille: https://Petra-Wille.com
Mentioned in this episode:
- Claude Code
- Anthropic
- ChatGPT
- Teresa’s blog post, Claude Code: What It Is, How It's Different, and Why Non-Technical People Should Use It
- Claude Code /init for project setup
- Henrik Nyberg
- Claude Opus 4.5
- Teresa’s blog post, How to Use Claude Code: A Guide to Slash Commands, Agents, Skills, and Plug-Ins
- Trello
- Ghost
- Petra’s FREE Product Leadership Newsletter
- Petra’s book, Strong Product People
- Product at Heart videos
- Claude’s Web fetch tool
- Wisprflow
- Petra’s Linkedin
- Andy Crestodina
- Product Talk articles
- Previous ATP episode: Happy New Year! Where Teresa said that she is starting to dig into the research on synthetic users
- Zettelkasten
- John Dewey's How We Think
Join the Conversation:
Have thoughts on this episode? Leave a comment below.
Full Transcript
Full transcripts are only available for paid subscribers.