aidesignco

An independent design media company

The AI era has new interface patterns. We find them.

For product designers and design-engineers building AI products, aidesignco turns emerging interface behavior into evidence-graded patterns, prototypes and components they can use.

One email when new work ships: the pattern, the prototype, the evidence. We use your address for the newsletter — nothing else. Privacy.

What we do

Patterns decoded, prototyped, and graded.

01 — Decode

We take emerging AI interface patterns apart: what the interaction does, why it holds up, where it breaks. Then we name it, so your team can talk about it.

02 — Prototype

We rebuild the patterns worth keeping as working artifacts — states, failures and recoveries included — so you see the mechanics, not just the happy path.

03 — Grade

Every claim we publish carries a Reality Ledger record: what state the thing is in, and how we know. A tested behavior never dresses up as a company promise.

The questions we keep answering: which AI interface patterns actually work, and why · how to design agentic, probabilistic UX that people trust · how to ship AI UI with taste, not slop.

The 30-day promise

A month of following should change how you work.

  • Name You can identify the AI interface patterns you encounter and call them by a shared vocabulary.
  • Judge You can evaluate an AI interface — trust, handoff, recovery, uncertainty — against our frameworks.
  • Use You can put working components from The Kit to use in your own product work, as releases ship — each one running code plus its Figma counterpart, with maturity and limitations declared.

If a month of our output doesn't deliver that, the output — not the promise — is wrong. We're at the very start: the first decodes are in production and the first Kit component ships when it genuinely runs, not before. The newsletter is where each one lands first.

The formats

Three ways a pattern reaches you.

Pattern Lab

The spine: AI interface patterns beyond chat, decoded and prototyped, with agent-UX teardowns. Every pattern post ships a built artifact.

Slop Makeover

We de-slop an AI-generated screen — every decision explained. It only ships at our top grade; a below-grade week is a skipped week.

In the Wild

Spotlights of genuinely strong, newly shipped AI UI in real products — first-hand tested, never sponsored, always tied back to our framework.

See how each format works →

How we earn trust

Evidence you can audit, errors we correct in public.

Every public claim we make about an interface carries two labels: what state the thing is in (shipping, beta, prototype, concept, unverified) and how we know (we tested it ourselves, the company documented it, someone credible reported it). When we get something materially wrong, we say so — publicly, timestamped, and explained, within 72 hours of confirming it.

Newsletter

Sign up, get the AI Reality Map now.

Every model, priced and verified against the vendor’s own page — the PDF is yours the moment you subscribe, no confirmation hoops. Then one email when the desk publishes: the verdict, the receipts, the number that didn’t survive checking.

We use your address to send the newsletter — nothing else. Unsubscribe anytime. Privacy.