Feb 7
Loading video player...
srcs: https://x.com/KarelDoostrlnck/status/2019477361557926281?s=20 https://x.com/KarelDoostrlnck/status/2019477361557926281?s=20 https://x.com/gdb/status/2019566641491963946?s=20 https://x.com/MindBranches/status/2019578855950790833?s=20 https://x.com/buccocapital/status/2019598551228223526?s=20 How does OpenAI *actually* use Codex inside real workflows — and what does a $10k/month token budget really buy you? In this episode we break down: - How engineers reverse-engineer new Codex and SDK features before docs exist - Why token budgets are becoming a real leverage advantage - The simple tooling setups heavy users rely on (VS Code, shells, work trees) - How “agent captains” and shared skills files help teams scale AI usage - Why systems of record are becoming dumb pipes — and where the value is moving This isn’t a model comparison. It’s a look at **how serious teams learn faster and compound output with AI**. 00:00 Introduction and Overview 00:03 Codex vs. Opus: Initial Thoughts 00:28 Token Budgets and AI Leverage 01:54 Greg Brockman's Insights on AI Implementation 03:03 Creating Effective AI Agents 05:18 Scaling Research with Codex 07:00 System of Record and SaaS Commentary 08:54 SDK Workflow and Practical Tips 12:31 Conclusion and Viewer Interaction