Claude Haiku is the single biggest weakness in the Anthropic ecosystem now. Haiku is intended as a low cost, fast, and cheap model to do basic work – reading, summarization, web search, etc. And at the time of its release almost a year ago, it did those things reasonably well. But times change. Models change.…
When times are good, people buy ideas. When times are bad, people buy action. When times are good, people buy strategy. When times are bad, people buy tactics. When times are good, people buy brand. When times are bad, people buy price. When times are good, people buy growth. When times are bad, people buy…
Tell AI what wins. One of the things that will degrade AI performance the most – and silently – is conflicting instructions where the AI has to determine what you mean. If you look at the reasoning traces (internal chat) when you give conflicting instructions, you’ll see AI models literally chasing their tails. It looks…
Summary In today's episode, I walk through a three-layer framework for training large language models to capture a specific brand voice instead of sounding like generic corporate copy. Here's what this means for you. You gain a repeatable system that combines good and bad examples, prompts aligned to how models think, and quantitative QA loops…
This one hastily made slide explains pre-IPO jitters better than anything else when it comes to AI. DeepSeek released its v4 Flash model earlier this year, on April 24. At its release, it was widely regarded as pretty smart and stupid cheap for its size. For a lot of folks using agentic AI, its pay…
Summary In today's episode, I break down why most AI brand visibility tools are unreliable and what you should actually measure instead. Here's what this means for you. You can stop paying for snake oil dashboards and instead use free, vendor-sourced data to track how AI systems actually interact with your site. You'll also learn…
Summary In today's episode, I explain how to decide where to draw the line for human review in AI marketing tasks based on risk levels. Here's what this means for you. You'll calibrate your review process to match the consequences of each AI output so you scale low-risk work without exposing your brand to high-risk…
Summary In today's episode, I break down whether missing a micro trend actually costs marketers anything meaningful. Here's what this means for you. You'll see that chasing micro trends is really anomaly marketing, a costly attention treadmill that rarely produces durable brand value. You'll also learn these concepts: how your choice of audience and platform…
Last night I used Minimax M3 to generate a steaming pile of AI slop in 51 minutes, a 50,000 word 1920s romance novel set in Venice. (and it’s terrible) I did it to prove a point to a commenter on X who publicly posted they didn’t feel like human authors deserved fair compensation for their…
