I Built an Entire App with OpenAI's Codex and 8 AI Agent Employees
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
AI Agents are the new SaaS
Greg Isenberg
51.7k views
WTF Is an "AI Agent Loop"? Genius or Hype?
Greg Isenberg
69.3k views
The $1M+ Solo AI Agent Business (Full Course)
Greg Isenberg
153.0k views
Screensharing How to Start an AI Agent Business Today
Greg Isenberg
51.4k views
Making $$ with AI Agents
Greg Isenberg
102.0k views
Stop using Claude. Start using Codex?
Greg Isenberg
107.4k views
Hermes Agent: The New OpenClaw?
Greg Isenberg
136.7k views
How AI agents & Claude skills work (Clearly Explained)
Greg Isenberg
599.6k views
Paperclip: Hire AI Agents Like Employees (Live Demo)
Greg Isenberg
359.3k views
Building AI Agents that actually work (Full Course)
Greg Isenberg
532.2k views
Top Comments (10)
A few things i've experienced that kinda came to mind in this episode. 1) I no longer try to implement Supabase at all for an MVP. When first building a project, the database changes a lot, and I've found that supabase (especially things like RLS policies) start to become a mess and a lot of time is spent fixing database issues, so personally I would just build the mvp with sqlite or even json file databases because it's all controlled locally by the AI and it can easily make fixes on the fly vs sending you to supabase to get api keys and edit RLS policies and such. Plus, a lot of projects (especially if they don't have user accounts) don't even need something like supabase anyway, if you're just storing some admin fields then a built in tool is more than fine. 2) Whether its claude code and codex or claude and cursor or whatever, I found that doing every process with two AI's works great. One of them builds the plan, the 2nd one executes the plan.. use the plan builder to review the plan and issue fixes to the coder. I assume this is handled nowadays with agents but it's just what i've been doing because two different AI's also have two totally different perspectives. I also use two AI's to review a plan, or really review most everything, and then share the plans with each other to improve.
The concept of AI agents as team members is revolutionary! Seeing how Codex can help non-technical founders build MVPs this fast is a game-changer for the startup world.
This got me so fired up for accelerating my building. On a roll Greg, another great episode.
Awesome. I usually watch you guys separately. So This is like watching an episode of DC Comics Presents or Marvel Team-Up.
No fluff just pure idea and execution. Love it ❤
This is so cool, I will definitely build one app with CodeX.
dude i'm so impressed with your setup
I LEARNED PRD TODAY!!! THANK YOU FOR UP SKILLS YOUR THE BEST IN THIS SPACE G
Thank you *Greg Isenberg* and *Alex Finn* Definitely connected to the value of build fast and hidden benefits to releasing quickly.
I learn a lot ever since following this channel and I thought of many ideas as I watch Greg and his guests' demonstrating. This is a sauce.
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
A few things i've experienced that kinda came to mind in this episode. 1) I no longer try to implement Supabase at all for an MVP. When first building a project, the database changes a lot, and I've found that supabase (especially things like RLS policies) start to become a mess and a lot of time is spent fixing database issues, so personally I would just build the mvp with sqlite or even json file databases because it's all controlled locally by the AI and it can easily make fixes on the fly vs sending you to supabase to get api keys and edit RLS policies and such. Plus, a lot of projects (especially if they don't have user accounts) don't even need something like supabase anyway, if you're just storing some admin fields then a built in tool is more than fine. 2) Whether its claude code and codex or claude and cursor or whatever, I found that doing every process with two AI's works great. One of them builds the plan, the 2nd one executes the plan.. use the plan builder to review the plan and issue fixes to the coder. I assume this is handled nowadays with agents but it's just what i've been doing because two different AI's also have two totally different perspectives. I also use two AI's to review a plan, or really review most everything, and then share the plans with each other to improve.
The concept of AI agents as team members is revolutionary! Seeing how Codex can help non-technical founders build MVPs this fast is a game-changer for the startup world.
This got me so fired up for accelerating my building. On a roll Greg, another great episode.
Awesome. I usually watch you guys separately. So This is like watching an episode of DC Comics Presents or Marvel Team-Up.
No fluff just pure idea and execution. Love it ❤
This is so cool, I will definitely build one app with CodeX.
dude i'm so impressed with your setup
I LEARNED PRD TODAY!!! THANK YOU FOR UP SKILLS YOUR THE BEST IN THIS SPACE G
Thank you *Greg Isenberg* and *Alex Finn* Definitely connected to the value of build fast and hidden benefits to releasing quickly.
I learn a lot ever since following this channel and I thought of many ideas as I watch Greg and his guests' demonstrating. This is a sauce.