Anthropic publishes skills for Claude Code: structured task definitions that tell the model what to build and how. They're designed for Claude. I ran them on GPT-5.4 and Gemini 3.1 Pro too. Claude won. Claude Opus 4.6 averaged 0.825 across four Anthropic skills, against
Most companies treating agent-pickability as a distribution problem are solving the wrong layer. Getting into training data, appearing in MCP registries, reducing integration friction to five lines of code: that's necessary. It's also table stakes within eighteen months. The durable advantage is further down: whether your
Microsoft gave Office to students for nearly nothing, not as charity but as customer acquisition cost arbitrage. Four years of muscle memory did the selling, and enterprise licenses followed for three decades. Something structurally similar is happening to dev tools, except the conversion funnel has collapsed from years to minutes
Vibe coding gets you to a working prototype faster than anything before it. The part that trips teams up is the handoff. The pattern shows up consistently now. Someone spends a weekend prompting their way to a functional app. The demo lands well, leadership sees velocity, and engineering gets the
Howard Marks' new AI memo buries its most important observation in a quote from Claude: if 30-50% of structured knowledge work migrates to AI compute, that's $150-250B in annual labor value shifting. The market already priced this in, violently. The SaaSpocalypse erased roughly $1T in enterprise software
The fastest way to understand AI isn't to study it. It's to build something small that solves a problem you actually have, not a tutorial project or a course, but something you'll use tomorrow morning. A script that sorts your email backlog by urgency.
Last week I asked Claude to generate architecture documentation for a system I'd built. The system is called naalayiram — an AI-powered exploration platform for 4,000 ancient Tamil poems. Claude produced a 16-page PDF with embedded diagrams, then a comprehensive architecture PNG when I wasn't satisfied
Claude Sonnet 4.6 dropped on February 17th. By that evening I had run a full evaluation and decided to stay on 4.5. Over the next two days, I ran the same pipeline a dozen more times across both models. The answer didn't change. What I tested
The S&P 500 software sub-index is down 27% since October. Private equity firms are tripping over each other to downplay their software exposure. The Economist reports $500 billion in software-linked debt lurking in US credit markets. And "SaaS is dead" is having its moment as a
In November 2025, Google shipped Antigravity, a full IDE built around the Gemini 3 model. By January 2026, Anthropic had extended Claude Code's agentic capabilities into its desktop app. On February 2, 2026, OpenAI released the Codex desktop app for macOS. Three frontier labs, three consumer-facing coding applications,
TL;DR: If AI assistants can reinterpret product flaws as features based on your specific goals, that changes everything. When Gemini recommended socks that hikers panned (loose elastic is bad for hiking but good for circulation), it demonstrated what could be goal-aware inference. If this capability scales, it threatens the
When the Same Patterns Reveal Different Products I built two MCP servers to test the same question: should this product be agentic? Same patterns applied to both. One confirmed the agentic hypothesis. The other falsified it. What I'd been calling a "movie recommendation agent" was actually
AgenticAI
A Framework for Deciding If Your Product Should Be Agentic When someone brings me a product idea, my first questions are always the same: Who's the customer? What's the pain? Is there a market? Will they pay? Do that work first. No shortcuts. But once you&
AI
Here's something I've noticed after months of heavy use: Claude handles file operations with surprising reliability. ChatGPT... doesn't. Not consistently, anyway. I'm not talking about edge cases. I mean simple requests - "export this as markdown," "give me a
AI
The Problem "Which LLM is best?" is the wrong question. "Best for what?" is the right one. Traditional evals tell you which model is smartest. This tells you which model is right for your specific task. Every model has a design philosophy - a personality that
AgenticAI
The System Structure When The Economist declared AI’s war on the “rip-off economy,” they documented a transformation that’s been 50 years in the making. But the real story isn’t just about technology -it’s about feedback loops that have kept information asymmetries stable despite decades of attempted
AgenticAI
If you’ve been building AI agents lately, you’ve probably seen this pattern (pun intended): We’re in the middle of an Agentic AI gold rush. LangChain, AutoGen, CrewAI, and dozens of other frameworks promise to make building AI agents as easy as importing a library. And they do
SystemsThinking
How OpenAI rewired the economics of compute through two feedback loops I love seeing systems loops in action. Feedback mechanisms reveal how complex systems actually work. And right now, we're witnessing two of the most elegant — and opposite — capital loops ever engineered in tech In the span of