Lenny Rachitsky · 8 min read · Gen AI Tools
🎙️ How I AI: Claude Opus 5 Review + Browser use in Codex + How Cursor and a Raspberry Pi makes AI fun
Read the original on lennysnewsletter.com →
Claire demonstrates how to use Codex to automate browser tasks, test onboarding flows, manage LinkedIn, and even operate an iPhone via screen mirroring, illustrating practical AI-powered productivity workflows. The episode also covers when to scale compute, persona-based testing to surface real user friction, and strategies for handling obstacles such as CAPTCHA.
Key Takeaways
- AI testing can be far more exhaustive than human testing, uncovering blocking issues that slip through when testers only follow happy paths.
- Frontier models perform better when given room to think and decide how to approach a task, leading to broader coverage with fewer blind spots.
- Persona testing with browser use reveals friction that synthetic user research misses.
- Balance compute and task complexity by matching model effort to the job, using smaller models for routine work and human input for verification when websites block automation.
Excerpt
Computer and browser use in Codex (5 real examples) Listen now on YouTube • Spotify • Apple Podcasts Brought to you by: Runway —The creative AI platform for images, video, and more Hyperagent —Deploy fleets of agents that handle real work Claire shows how she uses Codex to control her browser, test apps, manage LinkedIn, and even shop for her—plus the simple prompting trick that makes computer use work better. Biggest takeaways: AI testing can be far more exhaustive than human testing. When Claire tests her own onboarding flow, she naturally follows the happy path. She fills out every required…