StarZero All shows Try the podcast clipper →

May 21, 2025 · 35m · startup-ideas

ChatGPT Codex is like 10 AI software developers (tutorial for beginners)

Ben Tossell · 25m spoken Greg Isenberg · 4m spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

Host Greg Isenberg and no-code authority Ben Tossell demonstrate how non-technical creators can leverage OpenAI's browser-based Codex and GitHub to build, test, and deploy real-world software applications. The tutorial covers foundational version control concepts, incremental product development, and multi-tool AI debugging strategies to make software engineering accessible to everyone.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Greg holds 13.7% of the talking time here. How this is scored →

Greg as informed peer 2.9 Guest teaching 5.6 Guest disagreement 1.1 Greg pushing back 1.3
05100:0010:0020:0030:000:44–5:53 · Greg as informed peer 2/10 Ben Tossell on Navigating Codex as Non-Technical Creator Greg introduces Ben and sets up a beginner-friendly frame, asking basic setup questions. Ben takes the lead explaining GitHub repositories, commits, and how he converted his Carrd site into code.5:54–10:52 · Greg as informed peer 1/10 Live Demonstration Prompting Codex and Submitting Pull Requests Greg steps aside to let Ben do a live terminal and prompt demonstration. Ben walks through executing a prompt, inspecting file changes, and creating a pull request.10:53–15:24 · Greg as informed peer 4/10 Comparing OpenAI Codex With Prototyping App Builders Greg challenges Ben on why a non-technical person would use Codex when prototyping tools like Bolt seem far easier. Ben answers bluntly that maybe they shouldn't, but explains that text-to-app builders build fragile illusions whereas Codex builds robust iterative changes.15:25–18:16 · Greg as informed peer 2/10 Managing Pull Requests Merging Changes and Metric Tracking Greg jokingly tests Ben on whether he understands the positive and negative diff numbers. Ben clarifies the diff metrics and demonstrates how closing PRs isolates failed experiments without breaking the main branch.18:17–21:54 · Greg as informed peer 2/10 Beginner Best Practices for Building and Safe Rollbacks Greg prompts Ben for best practices. Ben details how non-technical builders can safely rollback broken builds on GitHub and work around Codex's lack of UI upload tools by using v0.21:55–29:48 · Greg as informed peer 3/10 Effective Debugging Strategies and External Model Prompting Context Greg asks for clarification on using ChatGPT versus Codex ask modes, and questions why environment setups are unintuitive. Ben demonstrates advanced debugging workflows like uithub and setting up terminal dependencies for open-source repos.29:49–35:00 · Greg as informed peer 6/10 Final Takeaways on AI-Powered Development for Beginners Greg articulates a strong high-level synthesis, framing Codex as a lightweight gateway for learning computer science concepts and predicting UI directions for AI dev tools. Ben agrees and reinforces why bottom-up iterative building prevents AI debugging despair.0:44–5:53 · Guest teaching 5/10 Ben Tossell on Navigating Codex as Non-Technical Creator Greg introduces Ben and sets up a beginner-friendly frame, asking basic setup questions. Ben takes the lead explaining GitHub repositories, commits, and how he converted his Carrd site into code.5:54–10:52 · Guest teaching 5/10 Live Demonstration Prompting Codex and Submitting Pull Requests Greg steps aside to let Ben do a live terminal and prompt demonstration. Ben walks through executing a prompt, inspecting file changes, and creating a pull request.10:53–15:24 · Guest teaching 6/10 Comparing OpenAI Codex With Prototyping App Builders Greg challenges Ben on why a non-technical person would use Codex when prototyping tools like Bolt seem far easier. Ben answers bluntly that maybe they shouldn't, but explains that text-to-app builders build fragile illusions whereas Codex builds robust iterative changes.15:25–18:16 · Guest teaching 5/10 Managing Pull Requests Merging Changes and Metric Tracking Greg jokingly tests Ben on whether he understands the positive and negative diff numbers. Ben clarifies the diff metrics and demonstrates how closing PRs isolates failed experiments without breaking the main branch.18:17–21:54 · Guest teaching 6/10 Beginner Best Practices for Building and Safe Rollbacks Greg prompts Ben for best practices. Ben details how non-technical builders can safely rollback broken builds on GitHub and work around Codex's lack of UI upload tools by using v0.21:55–29:48 · Guest teaching 7/10 Effective Debugging Strategies and External Model Prompting Context Greg asks for clarification on using ChatGPT versus Codex ask modes, and questions why environment setups are unintuitive. Ben demonstrates advanced debugging workflows like uithub and setting up terminal dependencies for open-source repos.29:49–35:00 · Guest teaching 5/10 Final Takeaways on AI-Powered Development for Beginners Greg articulates a strong high-level synthesis, framing Codex as a lightweight gateway for learning computer science concepts and predicting UI directions for AI dev tools. Ben agrees and reinforces why bottom-up iterative building prevents AI debugging despair.0:44–5:53 · Guest disagreement 1/10 Ben Tossell on Navigating Codex as Non-Technical Creator Greg introduces Ben and sets up a beginner-friendly frame, asking basic setup questions. Ben takes the lead explaining GitHub repositories, commits, and how he converted his Carrd site into code.5:54–10:52 · Guest disagreement 0/10 Live Demonstration Prompting Codex and Submitting Pull Requests Greg steps aside to let Ben do a live terminal and prompt demonstration. Ben walks through executing a prompt, inspecting file changes, and creating a pull request.10:53–15:24 · Guest disagreement 3/10 Comparing OpenAI Codex With Prototyping App Builders Greg challenges Ben on why a non-technical person would use Codex when prototyping tools like Bolt seem far easier. Ben answers bluntly that maybe they shouldn't, but explains that text-to-app builders build fragile illusions whereas Codex builds robust iterative changes.15:25–18:16 · Guest disagreement 1/10 Managing Pull Requests Merging Changes and Metric Tracking Greg jokingly tests Ben on whether he understands the positive and negative diff numbers. Ben clarifies the diff metrics and demonstrates how closing PRs isolates failed experiments without breaking the main branch.18:17–21:54 · Guest disagreement 1/10 Beginner Best Practices for Building and Safe Rollbacks Greg prompts Ben for best practices. Ben details how non-technical builders can safely rollback broken builds on GitHub and work around Codex's lack of UI upload tools by using v0.21:55–29:48 · Guest disagreement 1/10 Effective Debugging Strategies and External Model Prompting Context Greg asks for clarification on using ChatGPT versus Codex ask modes, and questions why environment setups are unintuitive. Ben demonstrates advanced debugging workflows like uithub and setting up terminal dependencies for open-source repos.29:49–35:00 · Guest disagreement 1/10 Final Takeaways on AI-Powered Development for Beginners Greg articulates a strong high-level synthesis, framing Codex as a lightweight gateway for learning computer science concepts and predicting UI directions for AI dev tools. Ben agrees and reinforces why bottom-up iterative building prevents AI debugging despair.0:44–5:53 · Greg pushing back 1/10 Ben Tossell on Navigating Codex as Non-Technical Creator Greg introduces Ben and sets up a beginner-friendly frame, asking basic setup questions. Ben takes the lead explaining GitHub repositories, commits, and how he converted his Carrd site into code.5:54–10:52 · Greg pushing back 0/10 Live Demonstration Prompting Codex and Submitting Pull Requests Greg steps aside to let Ben do a live terminal and prompt demonstration. Ben walks through executing a prompt, inspecting file changes, and creating a pull request.10:53–15:24 · Greg pushing back 4/10 Comparing OpenAI Codex With Prototyping App Builders Greg challenges Ben on why a non-technical person would use Codex when prototyping tools like Bolt seem far easier. Ben answers bluntly that maybe they shouldn't, but explains that text-to-app builders build fragile illusions whereas Codex builds robust iterative changes.15:25–18:16 · Greg pushing back 1/10 Managing Pull Requests Merging Changes and Metric Tracking Greg jokingly tests Ben on whether he understands the positive and negative diff numbers. Ben clarifies the diff metrics and demonstrates how closing PRs isolates failed experiments without breaking the main branch.18:17–21:54 · Greg pushing back 0/10 Beginner Best Practices for Building and Safe Rollbacks Greg prompts Ben for best practices. Ben details how non-technical builders can safely rollback broken builds on GitHub and work around Codex's lack of UI upload tools by using v0.21:55–29:48 · Greg pushing back 2/10 Effective Debugging Strategies and External Model Prompting Context Greg asks for clarification on using ChatGPT versus Codex ask modes, and questions why environment setups are unintuitive. Ben demonstrates advanced debugging workflows like uithub and setting up terminal dependencies for open-source repos.29:49–35:00 · Greg pushing back 1/10 Final Takeaways on AI-Powered Development for Beginners Greg articulates a strong high-level synthesis, framing Codex as a lightweight gateway for learning computer science concepts and predicting UI directions for AI dev tools. Ben agrees and reinforces why bottom-up iterative building prevents AI debugging despair.

speaking balance: gold is Greg, purple is the guest (3 minute bins)

0:00 · Greg 45.6% · guest 54.4%0:00 · Greg 45.6% · guest 54.4%3:00 · Greg 2.9% · guest 97.1%3:00 · Greg 2.9% · guest 97.1%6:00 · Greg 0.5% · guest 99.5%6:00 · Greg 0.5% · guest 99.5%9:00 · Greg 16.6% · guest 83.4%9:00 · Greg 16.6% · guest 83.4%12:00 · Greg 0% · guest 100%12:00 · Greg 0% · guest 100%15:00 · Greg 8.8% · guest 91.2%15:00 · Greg 8.8% · guest 91.2%18:00 · Greg 8.5% · guest 91.5%18:00 · Greg 8.5% · guest 91.5%21:00 · Greg 14.8% · guest 85.2%21:00 · Greg 14.8% · guest 85.2%24:00 · Greg 2.8% · guest 97.2%24:00 · Greg 2.8% · guest 97.2%27:00 · Greg 9.6% · guest 90.4%27:00 · Greg 9.6% · guest 90.4%30:00 · Greg 43.5% · guest 56.5%30:00 · Greg 43.5% · guest 56.5%33:00 · Greg 11.3% · guest 88.7%33:00 · Greg 11.3% · guest 88.7%
Sharpest disagreement ▶ 11:10 Ben dismisses builder hype with 'maybe they shouldn't'

When asked why a non-technical beginner would choose Codex over friendly prototyping builders, Ben bluntly pushes back against conventional tech hype by admitting most beginners probably shouldn't use it yet.

 Hardest push from Greg ▶ 10:53 Greg confronts Codex's overwhelming UX

Greg directly challenges the premise of the demo, arguing that Codex looks 10x more intimidating than alternative tools like Bolt and Lovable.

 Biggest teaching moment ▶ 22:07 The uithub repo-to-token debugging hack

Ben teaches Greg an actionable debugging shortcut by replacing github with uithub to copy an entire codebase's token payload straight into ChatGPT.

 Greg holds their own ▶ 31:55 Greg maps out the future of task-based AI dev tools

Greg moves beyond the student role and demonstrates strong product foresight, explaining how task-based to-do list UIs will spread across competing platforms like Cursor and Lovable.
the scores for every segment, with the reasoning behind each
ChapterTopicGreg as informed peerGuest teachingGuest disagreementGreg pushing backWhy
Ben Tossell on Navigating Codex as Non-Technical Creator 2511 Greg introduces Ben and sets up a beginner-friendly frame, asking basic setup questions. Ben takes the lead explaining GitHub repositories, commits, and how he converted his Carrd site into code.
Live Demonstration Prompting Codex and Submitting Pull Requests 1500 Greg steps aside to let Ben do a live terminal and prompt demonstration. Ben walks through executing a prompt, inspecting file changes, and creating a pull request.
Comparing OpenAI Codex With Prototyping App Builders 4634 Greg challenges Ben on why a non-technical person would use Codex when prototyping tools like Bolt seem far easier. Ben answers bluntly that maybe they shouldn't, but explains that text-to-app builders build fragile illusions whereas Codex builds robust iterative changes.
Managing Pull Requests Merging Changes and Metric Tracking 2511 Greg jokingly tests Ben on whether he understands the positive and negative diff numbers. Ben clarifies the diff metrics and demonstrates how closing PRs isolates failed experiments without breaking the main branch.
Beginner Best Practices for Building and Safe Rollbacks 2610 Greg prompts Ben for best practices. Ben details how non-technical builders can safely rollback broken builds on GitHub and work around Codex's lack of UI upload tools by using v0.
Effective Debugging Strategies and External Model Prompting Context 3712 Greg asks for clarification on using ChatGPT versus Codex ask modes, and questions why environment setups are unintuitive. Ben demonstrates advanced debugging workflows like uithub and setting up terminal dependencies for open-source repos.
Final Takeaways on AI-Powered Development for Beginners 6511 Greg articulates a strong high-level synthesis, framing Codex as a lightweight gateway for learning computer science concepts and predicting UI directions for AI dev tools. Ben agrees and reinforces why bottom-up iterative building prevents AI debugging despair.

Statements from this episode (9)

Disclosure
Tossell used Factory AI to convert personal site code to GitHub
“I actually used a different coding agent factory to then say, put this all into code and put it on this GitHub repo. And that's where this then started off.”
Ben Tossell May 21, 2025 ▶ 5:14
Opinion
Text-to-app AI builders like Bolt create polished UIs with broken features
“I just think generally text to app builders are over building everything or completely setting everything up as an overbuilt tool. But actually all the features are under, under built. So if you ask for like a Shopify, a Spotify clone with Bolt, it'll spit out…”
Ben Tossell May 21, 2025 ▶ 13:02
Prediction Not checkable as stated
AI coding agents will eventually fully automate GitHub pull request workflows
“I just think you start here and you can think, well, at some point the, like the PR stuff, that's going to be automated. Like, I'm sure you can get bots now that if there's a successful pull request, it'll just merge it to the main branch. So like, you never h…”
Ben Tossell May 21, 2025 ▶ 14:55
Assertion Not checkable as stated
Tossell updated his live website via AI with five button clicks
“We've opened the task, we've merged it, we've closed it works on my site. I did no coding, and I did Maybe like four or five button pushes to do this.”
Ben Tossell May 21, 2025 ▶ 16:56
Assertion Supported
Tossell: Codex cannot process uploaded files, images, or UI mockups
“Because the thing you can't do in codecs, so there's some limitations here, obviously, like you can't upload any files. You can't upload any images. You can't say make the UI look like this.”
Ben Tossell May 21, 2025 ▶ 20:55
Assertion Supported
Tossell: Codex Ask Feature Lacks Live Internet Access
“And it can do it and it, but it's more connected to, it's not connected to the internet. So it's like, I'm just looking at your repo and answering based on what it has, what knowledge it has to give you that response.”
Ben Tossell May 21, 2025 ▶ 23:23
Opinion
MCP makes complex workflows like database management very easy in Cursor
“Then there's things like MCP, which are making a lot of this very, very easy in things like cursor.”
Ben Tossell May 21, 2025 ▶ 24:20
Prediction Held up
AI builders like Lovable and Cursor will adopt task-based UIs
“And the lovables and bolts and curses of the world are probably going to introduce a to-do list task-based UI in the future because it just kind of, it makes a lot of sense.”
Greg Isenberg May 21, 2025 ▶ 32:47
Insight
Task-based AI tools are better for beginners than text-to-app builders
“I kind of feel like this is a better place to start than a text app builder, because like I said, it's going to build a big fancy thing. And what you're going to end up in is like seeing a bunch of terminology code being written tests failing. They're all to d…”
Ben Tossell May 21, 2025 ▶ 34:12
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 300 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more