AI-NATIVE DESIGN · PRACTICED IN THE OPEN
Give agents the rules, not screenshots.
Open manifest →78 PATTERNS · LIVE IN BROWSERTry the behaviour
Live interaction patterns, running in your browser.
Open →33 READINGS · 5 TRACKSRead in order
Canonical readings for the field, with my take on each.
Open →9 TERRITORIES · 37 NODESSee the map
The field, mapped. Changed when the field changes.
Open →10 CASE STUDIES · EVIDENCE FIRSTJudge the proof
Six substantial projects - the craft applied.
Open ↓LIVE RANKINGS · CHECKED 12 SEP 2026Compare the models
The generative UI ranking, verbatim from the live boards.
Open →THE TOOL LANDSCAPEPick the tools
The AI-native design stack - what to use, and when.
Open →10 STEPS · BRIEF → SHIPPEDLearn the process
The manual for AI-native design - tools, templates, failure modes.
Open →12 COMPONENTS · 6 BUILD STEPSBuild the components
AI-native components - what they are, and how to create your own.
Open →THE PLAYBOOK · THE WHOLE JOB, IN THE OPEN
Direction. Craft. Judgment. Proof.
Published, not claimed.
Leading AI-native design means doing the whole job where people can check it: write the process, build the components, keep the judgment current, ship the proof. That is this site.
The 10-step process for AI-native design - brief to shipped, with 52 tools, 19 templates and 30 failure modes.
Open →CRAFTThe components12 AI-native components catalogued with the states to design - plus the 6-step guide to creating your own.
Open →DEPTHThe starter kit17 fill-in templates: behavior specs, grounding briefs, eval rubrics, failure taxonomies.
Open →JUDGMENTThe benchmarksEvery ranking the site tracks, clubbed on one page - read off the live boards, stamped with the day.
Open →PROOF OF PACEThe watchThe field watched daily. Finds land only after they verify - 25 so far, each with its source.
Open →Reference, skills, and systems.
Design the whole agent system.
Open reference →01 · HARNESSES & WORKFLOWSDesign the environment around the model.
Read chapter →SKILLS · SKILL.mdInstall Ayodhya's design skills.
Open skills →DESIGN SYSTEMS SHELFSystems I return to.
See favourites →AGENT DESIGN SKILLSSkills that make judgment portable.
See shelf →PEOPLE I FOLLOWDesigners who sharpen the work.
See people →FOR AGENTSRead the site as structured guidance.
Open llms.txt →MACHINE-READABLE SYSTEMStart with the manifest.
Open manifest →Compact published experiments, kept in proportion to their source material.
Orbit
An AI-native product concept where evidence, intent and human judgment stay visible.
Read compact case ↗AI-nativeSkipwait: Instant Work Platform
A compact Vibe Design experiment for an instant-work platform.
Read compact case ↗HealthcareTAT: 24 hours to realtime
A compact concept for more direct X-ray abnormality annotation.
Read compact case ↗- impact unclear
+ impact: 38% faster review
Awaiting review
Use this pattern +
Reviewed edit
↗Warning at 80% · oldest model is cleared first
Use this pattern +
Storage meter
↗Intent
Evidence
Exit condition
Use this pattern +
Attachment
↗✓Collect evidencedone
✓Check boundarydone
↻Prepare reviewactive
Use this pattern +
Task rows
↗NEWSLETTER · DAILY
AI design and AI creativity,
in your inbox daily.
What shipped, what scored, what is worth reading - one short email a day, drawn from the same field watch that runs this site. On Substack as AI for Creativity.
Issues land on Substack - free, one short email a day.