Exclusive: Meta and Microsoft are trying to wean employees off Anthropic’s Claude. Two of Anthropic’s biggest corporate customers are pushing staff toward their own AI tools as costs climb—and competition intensifies. Full story: thein.fo/4AOdpRl

New legal startup Teddy AI raised $60 million today. Many basic details about the business, including the specific area of law it operates in and the identities of key executives, remain confidential. law360.com/pulse/legal-tech/…

They are on the iOS app — timekeeping is now so easy to automate. Claude pulls from all docs, meetings, AI working sessions, etc and generates a daily draft timesheet for me to edit/approve

There has never been a bigger gap between what is technologically possible and what the world understands to be possible. For AI-native builders, the world has fundamentally changed. Things that once required months of effort and millions of dollars can now be built by one or two talented people in hours or days. Business leaders outside the AI bubble know that AI can radically change their business, from how they run it to what they offer their customers, but they have no idea where to begin, or even who to trust to help them find out. I don’t blame them. AI is accelerating so fast right now that if you aren’t actively building and learning at the frontier, your entire strategy is almost certainly built on wrong assumptions. To use an analogy from history, right now some people are flying jets across the country while others are still dying of dysentery on the Oregon Trail. It doesn’t have to be that way. Today we’re launching Foundations Design & Engineering (FDE) to bridge that gap. Over the past two years, Foundations has grown into a community of 300+ committed AI builders across Seattle and SF. Increasingly, the scarce resource for many of them isn’t the ability to build. It’s finding great problems worth solving. FDE is our attempt to connect these two worlds. For existing companies eager to embrace AI, FDE should feel like magic. You express a business problem in plain English. We analyze your problem through an AI-native lens, assign the right team in our community to solve it, and help you get to the best outcome for your business as quickly and cheaply as possible. No armies of junior consultants. No incentive to maximize billable hours. Just founder-grade technologists applying hard-fought lessons from the edge of what’s possible today. Sometimes the answer is a few hours of expert help or a product that already exists. Sometimes it’s a full build. Sometimes it’s “the tech isn’t there yet, let’s find another way”. You have problems. We have builders looking for them. Let’s make the connection. fde.fndtns.org/

Law professors often ask students, "where do you draw the line?" Here's what they're really asking, and some thoughts for students on how to respond. papers.ssrn.com/sol3/papers.…

I am can hold my own in debates with the author of the leading treatise in my specialty. (And have before on facemain!) I get 10/10 no notes annual performance reviews (in biglaw). It’s not just the bottom quintile (heh). For $100/mo, I’ve been able to clone myself. At least on legal research, the AI generates the work product I would have a year or two ago. And in a tenth to a hundredth of the time. AI is ‘replacing’ Top Quintile Elite Human Capital©™ too.

I have been arguing this since early this year. Firms should try tools that actually work well instead of those with the shiniest marketing. Unfortunately many firms who have already bought into certain vendors are reluctant to explore and implement new tools. This is due to: (1) institutional inertia caused by the billable hour that disincentivises actual productivity increase; and (2) the amount of institutional weight that was put behind implementation of the tools the firms bought into. No one wants to admit that there has been no effective productivity increase (as is clear from the perspective of clients who have neither seen a change in billing structure nor the hours charged to them) and that some change is needed.

Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback. Here’s a look at the benchmarks:

12 years ago, @paulg published an essay on the importance of being kind. His point: to succeed in the startup world, you must not be mean. Mean people fail. His reasoning: 1. Mean people fight, and fighting is stupid. It's a poor use of your mental and physical resources. 2. Mean people have insufficient networks - they can't get the best people to work with them. 3. Mean people are usually extrinsically motivated while kinda people are intrinsically motivated. And the individuals who are motivated by more than money typically keep going longer than others. Loved this piece.

It's strange how much easier it gets not to make certain mistakes after making them just once. You'd think a mistake that was easy to avoid making twice wouldn't be too too difficult to avoid making even once, but that's not usually how it works.

My guy is really arming up for our podcast showdown

US appeals court upholds Thomson Reuters' landmark win in AI training lawsuit reuters.com/business/media-t…