We use analytics and advertising tools by default. You can update this anytime.
Manage optional tracking categories. Necessary cookies stay on so the site can function.
Staff Writer & AI Editorial Lead
Katie Parrott is a staff writer. She writes Working Overtime and contributes to Vibe Checks, Source Code, and Context Window.
A million words of failed drafts taught me that improving my AI setup sometimes means giving it fewer rules to follow
Anthropic's new model is a more capable partner than Sonnet 5, so long as you keep a hand on the wheel and know when to call in Opus
Fable-level power at roughly 60 percent less than Fable’s token price makes Opus 5.5 impossible to ignore
Plus: 13 beliefs about writing with AI, and an antivirus for the AI age
OpenAI’s new model impressed us with writing, operating software, and visual design. Anthropic’s Fable still has better instincts for building a product.
Use AI to make better editorial decisions, and carry what you learn into the next story
Our tests show where it replaces Fable and Opus—and where we’re keeping the alternatives
New apps, tools, and setup—here’s what we changed and why
Plus: the number KateBench taught us to count, and a six-agent solar crew helping decide whether to run the dryer
A new lab says the sameness of AI prose is a training problem. It made the writing less predictable—but not necessarily any better.
Off-the-shelf AI couldn’t meet Headway’s security and workflow needs, so it built its own. Now the whole company uses it.
Plus: An AI writing policy that makes writers do the thinking, Zuckerberg’s AI Future for Everyone runs on Meta, and a tool that keeps your agents from timing out
The app worked. Nobody, including me, had checked whether it was safe.
Plus: The race for AI coding supremacy, the models Every is using, and a one-shot data visualization
Plus: How to audit old agent instructions, habits you unlearn when you start working for yourself, a solution for Claude-isms, and the new-model tests we may be neglecting
Anthropic’s latest Opus can do impressive work—but getting it there may require dismantling the systems you already use
These four questions helped me decide which ones to keep, redesign, revisit, or retire
Power users revolt at OpenAI’s tool combination. Plus: The battle for knowledge-work OS, managing agents with Fable, and a 53 percent error-rate drop
Plus: A manual for handing recurring work to Opus, a skill for judging new tools in context, and the specialist model that beat the frontier for less
Anthropic’s new model can draft, code, and analyze competently, but every use case has a cheaper, faster, or smarter alternative
Katie Parrott
Dan Shipper