P
PROMPT_ENG
BONUS

BONUS INTERVIEWS

Conversations with journalists and media professionals who are adopting AI coding tools in their daily work. Each interview supplements a specific module with a real-world perspective.

A1a MODULE 1 22 min

Madi McCool

Grants & communications manager, NJ Civic Information Consortium

Madi recently transitioned from browser-based AI to Claude Code in the terminal. With no coding background, she used Claude and GitHub to rebuild her nonprofit's website — saving the cost of hiring a developer. This interview covers her learning process, how she picked up technical vocabulary by watching Claude work, and her honest take on the terminal vs. desktop app.

TOPICS COVERED

Background and role
Small nonprofit, wearing many hats, no coding experience
From writing help to website builder
Evolution from content drafting to full site rebuilds with Claude + GitHub
Learning by watching
Reading Claude's output as it works, absorbing patterns over time
Vocabulary as a superpower
Learning terms like SFTP, PRD, and defer — "cheat codes" for better prompts
Terminal vs. desktop app
When each makes sense, the terminal as "the older brother"
Setup walkthrough
Screen share: Claude Pro, Sonnet 4.6, Superpowers plugin, accept edits mode

KEY TAKEAWAYS

  • › You don't need a coding background to use CLI tools effectively. Madi went from zero coding knowledge to rebuilding a nonprofit website.
  • › Technical vocabulary is your biggest lever. Learning the right terms (SFTP, PRD, defer) lets you skip lengthy descriptions and give Claude direct instructions.
  • › It's not all-or-nothing. Madi still switches between the desktop app and the terminal depending on the task — and that's fine.
  • › "Your computer is not going to start smoking." Go in with intention, use the permission safeguards, and you won't break anything.

TIMESTAMPS

00:00Introduction
01:23Madi's role at the NJ Civic Info Consortium
01:57Early AI adoption and evolution of use
03:25Coding background (none) and learning curve
04:36Understanding GitHub — "Google Drive for code"
06:02Learning by reading Claude's output
09:44Technical vocabulary as "cheat codes"
11:36Terminal vs. desktop app — compare and contrast
15:06Screen share: Madi's setup and plugins
22:02Closing advice for beginners
A1b MODULE 1 26 min

Mike Janssen

Digital editor, Current.org

Mike uses Codex in VS Code to build automated workflows for his public broadcasting newsroom — with no prior coding experience. He automated his publication's newsletter, built systems to track nonprofit 990 filings and federal audits, and is developing an internal chatbot for reporters to query their archives. This interview covers how he got started, his "if the code works, it works" philosophy, and what he wishes he'd known earlier.

TOPICS COVERED

Background and role
Digital editor at a public broadcasting trade publication, no coding background
Newsletter automation
MailChimp HTML generation sorted by Parsley analytics, plus a Slack integration
Document tracking
Automated monitoring of nonprofit 990s, state budgets, and federal audits
"If the code works, it works"
Code is verifiable in ways text generation isn't — but watch for security blind spots
VS Code setup walkthrough
Screen share: Codex chat panel, terminal, file explorer layout
Context files and agents.md
Discovering and updating the project memory file that persists across sessions

KEY TAKEAWAYS

  • › You don't need coding experience to build useful newsroom automations. Mike went from zero to newsletter automation, document tracking, and internal research tools.
  • › Start small and practical. Mike's first project (newsletter automation) was a clear time-saver that took a day to build and cut a 30-minute task to 2 minutes.
  • › The AI itself can help you find your next project. Ask it what would be useful given your beat, and it can prioritize ideas by difficulty and value.
  • › Keep your context files updated. Mike had created an agents.md and forgotten about it — missing out on persistent session memory that prevents repeated mistakes.

TIMESTAMPS

00:00Introduction
00:53Mike's role at Current.org
01:40Getting started with Codex via Dolzan's newsletter
02:40No coding background — "almost entirely vibe coding"
06:03Newsletter automation with MailChimp + Parsley
07:36990 tracking, state budgets, federal audits
10:56"If the code works, it works" — verifiability vs. text generation
12:01Internal archive chatbot and financial document analysis
14:56Screen share: VS Code + Codex layout
16:12Context files and agents.md
20:03Pain points — the "bad golf swing" analogy
23:21Closing advice: start small, ask the AI for project ideas
More interviews coming soon

Each module will have bonus interviews with people using these tools in real newsrooms and media organizations.