Month: July 2026

  • Is the Next Generation of AI Already Here? How to Build Your Own Fable Ensemble

    What if Fable is an ensemble in a trenchcoat? Paul Roetzer recently commented that a fifth-generation Mythos/Fable open weights model is only 6-12 months away, regardless of government stipulations. That got me to thinking – you never, ever see the back end infrastructure of Claude or any closed-weights provider. Those of us in the open…

    Continue reading →

  • You Ask, I Answer: Building Audit Trails For Enterprise AI Agents?

    Summary In today's episode, I explain why enterprise AI agents are failing observability audits and how to fix it by treating them as ordinary software. Here's what this means for you. You'll learn to reuse the monitoring and logging infrastructure you already own to build verifiable audit trails for regulated industries. You'll also learn these…

    Continue reading →

  • Stop Wasting Money on AI: The Ultimate Guide to Choosing the Right Model for Every Task

    Use the right model for the right task to save token budget. Planning? Use the biggest, smartest models. Execution? Use a midrange model that can follow directions well. Reading and summarization? Use a light, fast, dumb model that does well when it’s given all the data. If you leave your favorite AI tool set on…

    Continue reading →

  • You Ask, I Answer: Managing Enterprise Shadow AI?

    Summary In today's episode, I answer a listener's question about how to track and manage shadow AI and unauthorized local model usage by employees in large enterprises. Here's what this means for you. You can turn what looks like a security loophole into a budget-saving strategy by governing local models instead of banning them outright.…

    Continue reading →

  • Almost Timely News: 🗞️ How To Do Feature Engineering with AI (2026-07-12)

    _Almost Timely News: 🗞️ How To Do Feature Engineering with AI (2026-07-12) :: View in Browser_ The Big Plug 👉 Please take my 6-question Reader Survey to tell me what you want! Content Authenticity Statement 95% of this week’s newsletter was made by me, the human. The glossary at the end I made with Google’s…

    Continue reading →

  • Optimization vs. Innovation: Why the Next Big AI Leap Isn’t in the Models Themselves

    “(Model) can’t even tell when to go to the car wash!” “(Model) reduced my workload by 84 years and built three new companies in an evening!” These are the two general hot takes on any new AI advancement, especially new models. They’re overly reductionist, absurd caricatures that don’t help us understand a new model and…

    Continue reading →

  • So What? Setting Up AI Bot Tracking in Google Analytics

    Summary In today's episode, I walk through the full setup process for tracking AI bot traffic in Google Analytics using Cloudflare Workers as the gatekeeper. Here's what this means for you. You gain a way to finally distinguish machine visitors from human visitors so your analytics reflect reality instead of inflated numbers. You'll also learn…

    Continue reading →

  • Stop Measuring the Wrong Metrics: Why ROI is Failing Your AI Strategy

    (Earned – Spent) / Spent = ROI A brief reminder that this is ROI, and nothing else is. Earned minus spent, divided by spent, is return on investment. Nothing else is ROI. If you do not know: What you earned What you spent You cannot measure ROI. Period, end of story. Lots of people are…

    Continue reading →

  • You Ask, I Answer: Enterprise AI Compute Costs?

    Summary In today's episode, I explain why hyperscalers are hoarding gigawatt-scale power capacity and what that power grab means for your enterprise AI costs. Here's what this means for you. You can escape rising token prices and cloud dependency by combining on-premise AI hardware with your own distributed green power generation. You'll also learn these…

    Continue reading →

  • Why You Must Use Lateral Thinking to Prevent AI From Stifling Your Creativity

    AI generally doesn’t think out of domain. There’s a technique in metacognition called lateral thinking, where you take a concept from one domain and you move it to another domain. For example, if you play World of Warcraft and you know how to win PvP at Arathi Basin, those same strategies and tactics can be…

    Continue reading →