On June 3, 2026, Matthew Prince, the CEO of Cloudflare, posted a number from his own network’s live data. For the first time in the history of the internet, more than half of all web traffic was not human. Bots, crawlers, and AI agents made up about 57 percent of the requests. People made up the rest. Prince said he’d figured the crossover would come at the end of 2027. Agentic traffic got there a year and a half early. ...
Grafe Auction runs about 290 auctions a year across 48 states. Each campaign takes 10 to 14 days from start to finish. That’s roughly one new marketing campaign going live every business day. Ad copy, search keywords, geographic targeting, photo selection, platform-specific formatting. Every time. For years, each campaign started the same way: a blank page. Someone opens the lot catalog, scans through it, figures out what’s worth highlighting, writes the ad copy, picks the photos, formats everything for Facebook and Google, selects the target markets. Then does it again tomorrow. And the day after that. ...
I was working through the details of bringing a small team in-house. Purchase price, salaries, org chart, who reports to whom. Normal acquisition stuff. And I caught myself thinking about it like a hiring decision. How many people, at what cost, doing what work. But we weren’t hiring. We were acquiring. The distinction matters, and I think it matters more in the auction industry than most people realize. When you hire someone, you’re renting their skills for as long as they stay. The moment they leave, whatever wasn’t documented or embedded in a system walks out with them. That’s not cynical, it’s structural. I wrote about this a few weeks ago in a different context: the relationships someone builds on your platform become yours, but the person is always rented. Everyone is. ...
The Observation I’ve been keeping notes in this vault for two years. Hundreds of calls, meetings, emails, references, thought cards. All wiki-linked, all tagged, all searchable. I thought the value was in the search. Find the note, read the context, prepare for the call. Then we added community detection. A clustering algorithm that groups nodes by edge density. Nothing fancy. Runs in 50 milliseconds. And suddenly I could see that my vault has 32 natural communities. Association operations is one cluster. State chapters are another. Client work is a third. Thought cards live in their own world, almost entirely disconnected from the operational clusters where the ideas actually apply. ...
Definition The agentic shift is not an upgrade to existing infrastructure. It replaces the assumptions that infrastructure was built on. “Each level isn’t defined by what the AI starts doing. It’s defined by what the human stops doing.” – Five Levels of Agentic Commerce “The cost of modern convenience is an involuntary autobiography that someone else owns.” – Digital Exhaust “The agentic web is the marketplace trap with different actors.” – Two Trust Systems ...
The Idea This came up while fixing a bug where stale cached data was silently injecting garbage into every outbound email for a full day. The cache never questioned itself. It just kept serving bad data. Human memory has a built-in half-life for each memory. The things you use stay strong. The things you don’t fade. That’s not a bug, it’s what keeps the signal-to-noise ratio manageable. You naturally surface what’s relevant because relevance reinforces the memory. ...
I told someone last week that his business couldn’t survive him. He’d called to pick my brain. Decades in the auction industry, multiple divisions, good revenue. But every decision in the company routed through him. Sales, operations, finance, client relationships. He’d spent his career being the best person on the microphone, and the rest of the business just happened around that skill. There were no documented processes. No decision-making frameworks anyone else could follow. If his potential successor took over tomorrow, they’d fail. Not because they’re incapable, but because the architecture underneath was never built for anyone else to run. ...
Dave Plummer is a retired Microsoft engineer. He created Windows Task Manager, ported Space Cadet Pinball to Windows, built the ZIP folder support, and wrote the Windows activation system for XP. He’s been coding since the TRS-80 days. Dropped out of high school. Worked at 7-Eleven. Went back to school, cold-emailed his way into Microsoft from Saskatchewan, and stayed for years building tools that billions of people still use daily. He now runs a YouTube channel called Dave’s Garage. Lex Fridman had him on episode #479, and four things from their conversation stuck with me. ...
I was listening to Acquired’s Google episode this morning, the Alphabet deep dive, and one detail stuck with me. Google launched Google Video before YouTube even existed. Google had the engineering talent, the infrastructure, the money. YouTube had nothing except a webcam, a website, and almost zero friction. YouTube won because anyone could upload a video in minutes. Google Video required you to fill out forms, wait for approval, deal with format restrictions. Same basic product, totally different levels of friction. The low-friction version attracted the creators, and the creators attracted the audience, and the audience attracted more creators. Users did the scaling. Google just had to buy it afterward. ...
Grafe Auction runs about 290 auctions a year across 48 states. Each campaign takes 10 to 14 days from start to finish. I’m the partner responsible for marketing, technology, and operations. That pace doesn’t leave much room for a system that’s clever but doesn’t hold up under pressure. It also doesn’t leave room for losing a thought from January that turns out to be the missing piece in February. The personal knowledge management crowd has an orthodoxy right now: tools don’t matter, just write. The AI crowd has its own: throw everything at the model and let it figure it out. Both are half right. Tools alone don’t create insight. Models alone can’t connect what they’ve never been given. ...