Worth your time?
Learn 99% of ChatGPT Work in 61 Minutes (Codex for Work) by Riley Brown
Watch on YouTube
This video mainly teaches: a long, hands-on walkthrough of GPT Work/Codex features with practical demos and little selling.
Why this score
- 4:38The host shows a generated 19-slide presentation and says the agent searched 87 websites and ran for 13:33, which supports high substance and on-screen evidence (raises substance and proof).
- 12:19Live Gmail plugin demo where the agent drafts, schedules, and shows the sent email to Emily, which is a verifiable, reproducible step that increases proof and usable material.
- 25:43Demo of deploying a site to Vercel with a live verscel.app link, showing end-to-end workflow (build, deploy, add domain) and boosting proof and substance.
- 41:01Agent drafts a reply claiming a price of "$250 million per video," an exaggerated/unverifiable income claim that raises claimInflation slightly.
The short version
Riley Brown demos ChatGPT Work (the merged Codex/ChatGPT agent environment) and walks through 14 core capabilities: creating polished documents, plugins, editable blocks and diagrams, hosted sites, image editing, branching chats, voice/remote control, in-app browsing, skills, automations, spreadsheets, and multi-agent workflows. He shows real examples, a 19-slide McKenzie-style deck built from web research, Gmail/Notion integrations, inline image edits, and spinning up work sessions from your phone, and explains when to use cloud vs local modes and plugins vs skills.
What it teaches
- Treat agents as tool-enabled coworkers: build a prompt that states the desired outcome, let the agent select tools (plugins/skills), and iterate over files it creates. Apply this when you need repeatable, multistep outputs (reports, sites, spreadsheets). Don't use it when legal/privacy constraints forbid granting tool
- Choose cloud vs local based on persistence and access: use cloud GPT Work for scheduled automations, cross-device continuation, and phone-driven workflows; use local/Codeex when you need desktop-file control or terminal commands. Apply cloud for background tasks that must run when your machine is off. The common mistak
- Separate plugins from skills: plugins are authorized app integrations (Gmail, Notion, Vercel); skills are small workflows you author and may be cloud or local. Create cloud skills for mobile/web use and local skills for desktop-only file operations. The common mistake is creating a skill in the wrong scope and expectin
Key moments
- 3:19GPT Work generates a 19-slide McKenzie-style presentation after searching 87 websites and ~13 minutes of work, showing agent-led research-to-deck automation
- 11:06Demonstration of the Gmail plugin: GPT Work reads recent emails and drafts a reply, then sends it, proving full read/write integration
- 20:25Mermaid diagram renders inline (Boom moment), illustrating how blocks convert Mermaid code to interactive diagrams inside the chat
- 25:43Deploying a GPT-generated site to Vercel directly from GPT Work, proving end-to-end build and public hosting via plugins
- 28:06Inline image edits and variant generation using GPT Image 2, plus commenting to make iterative visual tweaks
- 37:23Voice mode live demo: conversational control over Notion and live edits in the desktop app (shows voice controlling context and edits)
Quotes
- Specifically, we're talking about ChatGBT work, OpenAI's new agent tool, which is like Claude Co-work, but in my opinion, much better and easier to use on mobile, web, desktop, or any other platform.0:13
- It was the first time anyone made an AI chatbot accessible to the world.2:00
- However, the entire world opens up to you once you understand the power of plugins.9:00
- This is a completely new technology. I'm telling you, this is one of the most fun things you'll ever do.43:36
Skip the video. Keep the lesson.
More scored videos