Most people start their AI journey by asking how to save time. That is not a wrong question — but Anthropic's latest research, based on open-ended interviews with over 81,000 Claude users across 159 countries, suggests it may not be the most important one.

The most commonly reported productivity gain in the study was not speed. It was scope. Not doing existing work faster, but doing things that you simply couldn't before, because of budget, skills, or just the assumption that certain capabilities belonged to someone else.

This episode is about the difference between saving time with the boring middle and asking what is now possible that wasn't before, and why that second question is where the real opportunity lies.

What You'll Learn

Why the Anthropic study's methodology is unusual
The difference between efficiency gains and capability gains
How to identify your "boring middle" and what to do once you have sorted it
How to prevent your freed-up time from get absorbed back into more of the same
How a delivery driver and landscape gardener from the illustrate capability gains
What the Pocket OS incident reveals about AI agent permissions, and the simple rule that would have prevented it

Timestamps:

00:00 Introduction and Travel Plans

01:37 About the Anthropic Research

03:37 How the Study Actually Worked

07:36 The Headline Productivity Stat

10:35 The Four Types of Productivity Gain from AI

11:54 What the Data Says About Job Displacement

13:32 The Efficiency Game: What It Gets You and What It Misses

16:45 Why Automating the Wrong Things Makes You Faster at the Wrong Things

21:27 The Boring Middle: Why Consistency Is the Point

25:00 Capability Gains: Doing Things That Were Previously Off the Table

28:54 The Wrong Question: Efficiency vs. Capability