Turning vague content guidelines into measurable AI-ready standards
UNDRR's AI assistant couldn't learn from vague guidelines like 'keep it concise.' I extracted patterns from high-quality examples to create measurable standards.
All writing
1.0Everything
Digest videos into sketchnotes through AI
Daniel Vaughan built an agent that generates sketchnotes from YouTube videos — a visual triage step before committing to long-form content.
Large projects need large communicating
Ben Kuhn shares his playbook for managing complex projects at scale, from maintaining detailed plans to delegating management itself.
Making sense of Microsoft’s many AI Copilots
Five Copilots, one brand — and one table to explain what each does.
Measuring success beyond the page view
Analytics on page views and bounce rates won't tell you if your content has impact — here's how I measure what actually matters.
Creating warm, dappled light with CSS and SVG
I added a sunlit ambience effect to this site using pure CSS and SVG — no JavaScript, just gradients and blend modes.
Why AI can't be debugged like traditional software
Boyd Kane explains why treating AI like debuggable code misses the fundamental difference in how these systems work.
Plenty to learn from Fedora's approach to AI code
The first community-driven policy I've seen that treats AI as a tool requiring human accountability, not a shortcut.
Introducing a "Work" section
A home for evidence-led impact stories documenting measurable outcomes from my work on platforms, performance, and design systems.
Making disaster economics compelling for global policymakers
UNDRR's flagship disaster risk report needed to communicate $2.3 trillion in annual losses — without burying the message in data or slowing access for users in disaster-prone regions.
It's true, short disruptions costs lots of time
Meeting creep, quick questions and context switching crush flow.
Output: An HTML-native tag once again saves JS befuddling
Use the native HTML output element to announce dynamic results — semantic, accessible, and simple.
Context engineering beats prompt tricks for AI agents
Anthropic argues that “context engineering” — not prompt tricks — is how agents stay accurate at long horizons.
Developers are becoming context technical writers for AI
Fabrizio Ferri Benedetti highlights docs-driven development and the rise of technical writers as “context curators” so AI can truly RTFM.
Embracing px and rem for CSS
Miriam Suzanne from OddBird explains how combining CSS units can lead to more robust and user-friendly typography.