Don't Worry About the Vase Podcast GPT Agent Is Standing By
Jul 23, 2025
Discover the latest enhancements of OpenAI's offerings, including an exciting new agent mode designed for professional users. Learn how ChatGPT acts as an autonomous agent, streamlining tasks like scheduling and financial modeling with surprising accuracy. Hear real subscriber experiences that highlight both the promise and limitations of AI personal assistants. Lastly, delve into the vital safety measures for high-capability models, addressing biological risks and user frustrations, and explore the importance of understanding AI's capabilities for effective use.
AI Snips
Chapters
Transcript
Episode notes
Agent Handles Secure Logins
- Alex West successfully used ChatGPT Agent to log into protected accounts and remain logged in across sessions.
- It can access Cloudflare-protected sites though CAPTCHAs remain a barrier to full access.
AI Agents Could Revolutionize Permissions
- Current structures lack intermediate permission controls for assistants, leaving them all-powerful or useless.
- AI agents may change this by enabling controllable, safer delegation of tasks.
Agent as Helpful Intern
- Users report ChatGPT Agent as a reasonably good intern, automating dull tasks like comparison shopping.
- One user saved $80 by allowing it to comparison shop and add items to the cart for final approval.