Pat Ryan

My biggest takeaways from @bhalligan: 1. “When you have to eat a shit sandwich, don’t nibble.” Borrowed from Ruth Porat. The temptation during layoffs is to do a small one and grow into it, and then do another six months later. Just rip the bandaid off. 2. “If you want to kill a plant, have two people water it.” Have a DRI (directly responsible individual) for every problem you want solved. When you’re small, everyone’s in the room and aligned. At scale, no one knows what’s happening across org boundaries. Committees produce the same result as two people watering a plant: it either drowns or dies of neglect. 3. Never waste a good crisis. The best improvements at HubSpot came after major problems forced them to rethink processes. After a devastating outage in 2019, they completely overhauled their deployment approach, resulting in much higher quality and reliability going forward. 4. Hire the spicy candidate, not the safe consensus pick. HubSpot used to run eight-person panels. When half scored a candidate 4/4 and half scored 2/4, they’d pick the boring all-3s candidate instead. They flipped this—started choosing the polarizing people with real weaknesses—and their hit rate improved. They also shrunk panels from eight interviewers to four. 5. The biggest skill gap in young CEOs is being able to give hard feedback. Many Sequoia founders Brian coaches are in their late 20s, have never managed anyone, and now need to layer early employees or transition co-founders into different roles. These conversations are deeply unnatural. Brian’s solution: peer groups of about 15 CEOs at the same stage. Misery loves company, and they learn faster from each other than from a coach. 6. Build your team like the 2004 Red Sox—a mix of homegrown talent and strategic outside hires. Companies consistently underrate their internal talent and overrate fancy resumes from big companies. If it’s a close call between promoting internally or hiring externally, give your homegrown talent the chance. 7. Brian’s LOCKS framework for evaluating founders: Lovable (do people really, really want to work for them?), Obsessed (deeply passionate about the problem), Chip on their shoulder (almost all great CEOs have this), Knowledgeable (about their domain), and a Student (constantly learning, like an LLM). 8. AI will transform go-to-market, but not how you might think. Enterprise sales is the last white-collar job AI will replace—but the top of the funnel is about to flip. Buyers will start in ChatGPT or Claude instead of Google, making your website less important. Brian predicts every site will have an AI avatar on the homepage handling the first conversation, scored in your CRM, and then handed to a human rep who brings their own AI sales engineer to every call. 9. Starting a company has never been easier. Scaling one has never been harder. AI make building trivial, so the number of companies is mushrooming. But that means unprecedented competition and noise. The bottleneck becomes distribution. Technical founders can build but rarely know how to distribute. Brian says the fastest learners will win. And distribution is about to shift—getting found in ChatGPT and Gemini will matter as much as getting found in Google. 10. HubSpot was #1 on Glassdoor, which Brian now thinks was the wrong goal. Their employee NPS was 60; customer NPS was 25. They were spending half their management meetings on employee issues. Brian shifted the center of gravity: customer panels at every management and board meeting, and comp plans tied to retention and customer NPS instead of revenue. His takeaway: wanting to be liked isn’t a good feature of a CEO.

The golden age of discovery is here garryslist.org/posts/demis-ha…

We've added Plan Mode to Claude Code in Slack. When you give Claude a complex task it will ask you clarifying questions and show you an implementation plan before proceeding.

Cowork on Windows! Early days, we're actively developing, so expect a lot of improvements and changes. Grab it at claude.com/download and let me know what you think! Some nerdy below-the-hood technical details in thread →

Windows users, rejoice -- you can now try Cowork

Plus, you can now set global and folder instructions that Claude carries into every session. Cowork on Windows is in research preview and available to all paid Claude plans: claude.com/cowork

Cowork is now available on Windows. We’re bringing full feature parity with MacOS: file access, multi-step task execution, plugins, and MCP connectors.

NFL teams made $13.9 billion from national revenue (mostly media rights) in 2024 That's more than the entire NBA, MLB, NHL and MLS combined... by a lot

Northwestern is set to host Penn State at the new Ryan Field in its inaugural game on October 2 🏟️👀

With a week to reflect, I’m convinced this was the greatest moment in sports history. A 2-star recruit leads the worst program ever to a national championship, beats his hometown team that wouldn’t let him walk on, and finishes it by running over the same guy who knocked him out the year before, with his mother battling MS watching from the stands. If it was a Disney movie, you’d say it was too much.

Roger Penske: "This is really an amazing time for us as a company to support the 250th anniversary for the country, and there's no better way for us to bring automotive and speed into the D.C. area."

DoorDash founder Tony Xu: "You can't compete against an incumbent on their territory” Tony argues: “You have to find something where they're not incentivized to do it (Innovator's dilemma)… and you have to find an area where you think you can be advantaged.” For DoorDash, this was end-to-end delivery and focusing on suburbs rather than cities. The incumbents at the time didn’t want to touch end-to-end delivery because it was lower margin. Incumbents also focused on cities because of the network density, but DoorDash realized the market outside of city centers was actually the bigger opportunity because that’s where most people lived. “Knowing where the market is and knowing structurally why that's different and why that might be difficult for a competitor to serve, that's pretty important. Now, you also have to be correct on that bet… [Our bet on serving suburbs] turned out to be correct. But we didn't know that a priori.” Tony’s other piece of advice is that you have to be “super fast”. A key advantage versus incumbents is that they have to make capital allocation decisions across their many businesses. But you probably only have one product so you should be able to move much faster: “Focus is actually really easy… You’ve got to build that one product” Video source: @khoslaventures (2024)

.@bcherny on @gregisenberg pod: pca.st/episode/f4cde0…

THE ULTIMATE CLAWDBOT REPORT Every single important post here on X about @openclaw. Report and analysis by @blevlabs with the X API. docs.google.com/document/d/1Mz… It read 38,000 people in AI community here to learn about Clawdbot and its founder @steipete. This would not be possible if I hadn't built my lists of the entire AI community: x.com/scobleizer/lis… DOZENS of use cases, fun posts, tutorials, and more.

@openclaw @AlexFinn As Alex said, everyone thought this was 10–15 years away. Reality check: it’s already here. And it works. When CEOs and companies fully realize this, the shock to the global workforce will be brutal. More on CloudBot → clawd.bot clawd.bot

Clawd Bot feels like the "ChatGPT moment" for personal assistant AI, and it's about to add to the already high demand for RAM and CPUs on top of API AI data center usage. We are already in short supply in terms of memory, GPUs, CPUs, and advanced packaging, and here comes a on-device super cycle on top of the data center cycle. Besides the hardware upgrade cycle, Clawd Bot also demonstrates the value of chat surfaces like WhatsApp, Telegram, iMessage, etc., in the new AI world. Integrating personal assistants into chat surfaces feels like the killer use case. $META, $AAPL seem well-positioned. On that note, also kudos to Anthropic; they are taking the world by storm with Claude Code.

The most important companies that get built in a given year are rarely aligned to the trend that was "hottest." The deals that people were fighting for "access" to.

This year's national championship is the second-most-watched title game in the #CFBPlayoff era -- and the most-watched CFB game of any kind in 11 years. The game's viewership increased by 8 million from last season's Notre Dame-Ohio State matchup.

Indiana-Miami national title game got a 36% increase in viewership over last year's Ohio State-Notre Dame title game and was the most-watched CFB game since 2014-15.