I Spent $10,000 Testing AI Agents and Only These Survived

I Spent $10,000 Testing AI Agents and Only These Survived

More

Summary

Creator Riley Brown shares the AI agent stack he settled on after spending $10,000 testing different tools, and explains how he uses it for administrative work, running his business, and growing a following of more than two million people across platforms.

The video focuses on how he splits work between Claude Code and Codex, and why he says most people use them differently than he does. A major part is Claude Projects, which acts as a coordinator session that spins up both local sessions on his computer and cloud sessions. He demonstrates asking it to clip long-form videos locally, upload them to Drive, and build a shareable web artifact on short-form clipping best practices, then controlling it all from the Claude iOS app.

Brown also covers a video made with Claude Opus 5.5 that drew 160,000 views on Twitter, and previews V2, a free, open-source tool he describes as GitHub for video editing, which tracks prompts and files and lets teams fork and share videos across any editor or agent. He describes a tool at the center of his stack that makes it easier to switch between agent platforms and share plugins and skills. The video includes a sponsored segment from Superhuman.


📺 Source: Riley Brown · Published October 28, 2026
🏷️ Format: Comparison

1 Item

Channels

1 Item

Companies