The episode opened with the new Codex Micro device, a developer-focused keypad built for agentic coding workflows. The hosts discussed who the device is really for, whether it helps professional developers more than casual AI builders, and whether physical AI controls are a temporary bridge before voice and named subagents take over.

The middle of the episode moved into AI regulation and model strategy. The hosts compared China’s new restrictions on companion chatbots for minors with the lighter approach in the United States, then turned to Kimi Three, Thinking Machines Lab, Mira Murati, Inkling, Tinker, and the difference between open weight and open source models. The discussion focused on enterprise customization, whether foundation models matter more than frontier models in some business cases, and why a “not great yet” model may still be valuable if companies can train it for their own workflows.

The back half shifted into practical AI builds and robotics. Brian shared a personal face-measurement app built in Claude Code to track weight-loss changes from photos, Gareth described an AI DJ tool, Beth discussed a Cloud Code work board concept, and Andy compared Claude Code and Codex on project execution. The episode closed with robotics stories, including One X’s tendon-driven robot hand and San Diego researchers using tele-operated humanoid robots for live surgical procedures.

Key Points Discussed

00:00:18 Episode Intro And Hosts

00:01:27 Codex Micro And Think Louder

00:02:26 Micro As A Developer Tool

00:04:11 Voice Activation And Agent Controls

00:05:40 Carl Buys Micro For His Dev Team

00:07:01 Replaceable Keys And Programmable Controls

00:09:14 Stream Decks And Existing Shortcut Hardware

00:10:33 Micro As A Collector’s Item

00:11:04 Trigger Skills, PR Reviews And Reasoning Control

00:12:28 Who Is Codex Micro Actually For?

00:15:21 Hardware Controls Versus Voice Coding

00:17:25 Named Subagents Instead Of Manual Toggles

00:19:18 Work Boards And Agent Status Tracking

00:20:17 AI Regulation In China And The U.S.

00:20:46 Demis Hassabis And AI Safety Guidelines

00:21:13 China’s Restrictions On AI Companion Chatbots

00:23:44 Population, Fertility And AI Policy

00:24:28 Kimi Three Release Mention

00:24:43 Inkling And Thinking Machines Lab

00:25:28 Mira Murati Background

00:26:30 Inkling As An Open Weight Model

00:27:36 Foundation Models Versus Frontier Models

00:27:57 Tinker As The Customization Platform

00:28:25 Bridgewater Financial Reasoning Example

00:30:40 Tinker Predating Inkling

00:33:23 Enterprise Strategy For Open Weight Models

00:34:57 Ethan Mollick’s Early Inkling Reaction

00:36:15 Open Source Versus Open Weight

00:38:52 Model License Examples Across Providers

00:40:16 Thinking Machines’ Business Model

00:42:24 Brian’s Face-Tracking AI Build

00:44:05 Pupil Distance As A Measurement Anchor

00:45:19 Moving The Tool To Mobile Selfies

00:46:52 Gareth’s AI DJ Build

00:48:27 Beth’s Cloud Code Work Board Concept

00:50:00 Slash Goal, LFG And Session Limits

00:51:31 Fable Reset And Anthropic Credits

00:52:20 Codex Five-Hour Limit Removed

00:53:03 One X Robot Hand

00:54:11 Tendon-Driven Dexterity And Washable Hands

00:55:31 Tele-Operated Humanoid Robot Surgery

00:56:27 General Purpose Robots In Remote Surgery

00:57:11 Robots As Future Surgeons

00:58:47 Episode Wrap-Up

The Daily AI Show Co Hosts: Brian Maucere, Andy Halliday, Beth Lyons, Gareth.