ai.u crew discuss the shift from prompt-response AI chatbots to “AI coworkers” or computer-use agents that perform multi-step work across apps, highlighting Anthropic’s Claude Cowork ($20–$200/month), Microsoft Copilot Tasks ($30/user/month), and Perplexity Computer ($200/month). They describe the interaction change from asking questions to delegating outcomes, with humans increasingly acting as supervisors who define context, monitor progress, and apply judgment, while noting concerns that convenience may erode competence and that many workflows require undocumented institutional knowledge. They debate whether automating tasks is always worth the setup and trust costs, and suggest processes and software may need redesign. They also examine Anthropic’s qualitative study using an AI interviewer for 81,000 participants, weighing scale and multilingual benefits against lost human connection and empathy.

00:00 Welcome And Topic Shift

01:11 New Coworker Tools Overview

02:36 From Prompts To Delegation

04:41 Agency And Real Examples

08:22 Matt Wants Automation

10:24 Supervisor Mindset And Skills

14:28 Convenience Versus Competence

22:01 Three Lanes Of Coworkers

24:56 Token Spend And Real Debugging

29:26 Autopilot Limits And Hidden Knowledge

32:03 Tools Need Skill

33:08 Prompting Meets Expertise

35:44 Tribal Knowledge Problem

38:11 Is Automation Worth It

38:49 Trust And Context Costs

41:03 New Companies Advantage

42:00 AI As Flourishing Tool

44:31 Claude Interviews Study

48:57 What Humans Add

50:45 Where AI Fits Best

54:11 Human Connection Matters

56:51 Wrap Up And Feedback

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit aiunprompted.substack.com