Podcasts

OpenAI just put Codex into the ChatGPT mobile app. Google is pushing Gemini further into a true agent in the cloud and deeper into Android‑level experiences. It’s internally testing Remy, a “24/7 personal agent” that can take actions on your behalf (not public yet). Anthropic shipped Dispatch in March, turning Claude on your phone into a remote control for your desktop. 𝗪𝗵𝘆 𝗶𝘀 𝗲𝘃𝗲𝗿𝘆𝗼𝗻𝗲 𝗰𝗵𝗮𝘀𝗶𝗻𝗴 𝘁𝗵𝗶𝘀 𝗶𝗱𝗲𝗮? A big reason is that agents finally got go

Over the past month I have had a version of the same conversation with maybe twenty senior leaders. The titles vary. The company sizes vary. The industries vary. The questions do not. We have Copilot. But my team wants Claude or Cowork. Are they secretly using it? I need to show results this quarter. Where do I even begin? Vendors, consultants, and my team all say three different things. Who do I believe? I am being asked to innovate and cut costs. What do I prioritize? I don

Satya Nadella, on the launch of Microsoft's 2026 Work Trend Index yesterday noted that every firm will need to reconceptualize work as they build agentic systems. Every AI conversation is really an operating model conversation. As agents take on more of the execution, the opportunity in front of leaders is to expand and elevate human agency, and to redesign how work gets done around it. The data tells you exactly how far most firms are from capturing that opportunity. 1 in 5

Anthropic, Blackstone, Hellman & Friedman, and Goldman Sachs just launched a new AI-native enterprise services firm. It is backed by approx. $1.5B in committed capital, to embed Claude and Anthropic engineers inside private-equity-backed and mid-market companies. This is beyond a distribution play for Claude. The idea is to democratize access to "forward-deployed engineers." Here is what I believe the venture is really pricing in: 1. Forward-deployed engineering is the operat

At CRExCLT last week, I spoke to a room of executive leaders through that shift: from Industrial Age factories to AI Age cognitive factories. From linear planning to exponential change. From asking for "AI use cases" to realizing that every AI conversation is really an operating-model conversation. A few ideas that I shared: → Mediocrity is now effectively free. When average output is cheap and abundant, the only things that compound are quality, judgment, proprietary data, w

Most leaders I speak with can name three Western AI models. Very few can name three Chinese ones. That gap may start to cost teams real money this year. The MIT Technology Review reported earlier this year that Chinese open‑source models have surpassed US models in total downloads. The three names worth knowing: 1\ Kimi K2.6 Released April 20, 2026, this open model delivers coding performance in the same league as top Western systems at a much lower price. 2\ DeepSeek V4 Come

Clusters are won by regions that combine research depth, commercialization infrastructure, and workforce pipelines in the same ecosystem. Chicago is doing all three at once. IBM plans to create 750 AI, quantum, cybersecurity, and data-science jobs over the next five years at its new FutureNow Chicago center, supported by a $19M state tax credit and partnerships with the University of Chicago and University of Illinois Urbana-Champaign. That announcement sits inside a much lar

The biggest AI players are competing for your desktop. Three interesting moves this week: 1\ Amazon poured another $5B into Anthropic, deepening what was already one of the largest AI bets in history. 2\ Microsoft and OpenAI amended their exclusivity deal. Within 24 hours, AWS announced GPT-5.5, Codex, and OpenAI-powered agents on Bedrock alongside Claude, Llama, Mistral AI, and Amazon's own models. 3\ Andy Jassy posted yesterday about Amazon Quick, a desktop AI that connects

We don't program intelligence line by line anymore, we shape behavior through data, models, prompts, tools, and governance. But here's the flip side. AI is now shaping our behavior at work too and a new Microsoft Research study shows us where. Researchers analyzed 200,000 anonymized Bing Copilot conversations and mapped them to O*NET (the U.S. Department of Labor's database that breaks down 900+ occupations into the specific work activities they involve). That let them measur