Design & Craft

247 frameworks, rules, and principles from 10 practitioners — ready to drop into any AI agent.

Meng To, Jenny Wen, Karri Saarinen, Caitlin Kalinowski, Katie Dill, and 5 more

What's in this pack

28

frameworks

66

rules

78

heuristics

75

principles

Your AI can help you

  • Setting a quality bar a coding agent can actually apply
  • Deciding when polish is worth the shipping delay
  • Reviewing a UI that works but feels generic

See the difference

Generic AI gives you a checklist. Gavel gives you named experts who disagree — so you can decide who to follow.

Generic AI

"Our app works fine but it feels generic and cheap next to competitors. Where do we even start?"

Here are some things to consider:

  • Establish a consistent design system with reusable components
  • Improve typography, spacing, and color choices
  • Study competitor products for inspiration
  • Gather user feedback on the current experience
  • Consider hiring a designer or agency

The right choice depends on your specific situation...

Gavel
Gavel knows your context
Technical founder shipping AI-generated UI, no designer on the team

"Our app works fine but it feels generic and cheap next to competitors. Where do we even start?"

"Quality is a stack, and you cannot skip a layer. First utility, does it solve a real problem. Then usability, a chair you can sit in but that is extremely uncomfortable will not work out for you for very long. Craft and beauty come third, and they are not decoration: beauty signals care, and care is what earns trust."

Katie Dill

Utility, Usability, Craft

"Generic is what you get when nobody on the team has an opinion. Craft starts on the hiring front, with people who care, and it shows up as the absence of little paper cuts rather than as a redesign. Every one of those small annoyances is a reason someone stops using the thing."

Karri Saarinen

Opinionated Software

Where They Disagree

Dill runs quality as a system: a review cadence where the team walks its own product on a schedule and logs friction, so quality survives a growing org. Saarinen argues the cadence is downstream of taste, and that hiring people with opinions beats any process. Both reject the AB test as the arbiter, which is exactly where a generic assistant sends you.

See exactly what you get

Real items from this skill pack. Every item includes expert attribution and source material.

Framework

Block-Frame Conceptual Foundation

Block-Frame Conceptual Foundation Stay in ultra-low-fidelity block frames long enough to lock the conceptual model before any high-resolution visual work, so feedback stays on substance rather than surface. Steps: 1. Represent screens as big chunky blocks showing only location and hierarchy 2. Debate and refine the conceptual model while fidelity is too low for visual critique 3. Only after the foundation is solid move to wireframes then high-res comps 4. Leverage a robust design system so high-res comps can be produced in a day once locked Why it works: High-resolution artifacts immediately pull attention to colors, shapes and copy; low-fidelity forces the team to solve the hard thinking first and prevents premature commitment. Common mistakes: - Jumping straight to high-fidelity prototypes or GenAI comps - Letting PMs or execs see polished visuals before the concept is solid - Treating visual polish as the hard part instead of the conceptual heavy lifting

Bob Baxley

35 years of product design wisdom from Apple, Disney, Pinterest and beyond | Bob Baxley

single_practitioner consensus
Rule

Align all founders explicitly on the importance of quality and craft before hiring; multiple conf...

Align all founders explicitly on the importance of quality and craft before hiring; multiple conflicting cultures cannot coexist Context: Early company formation when founders come from different prior cultures

Karri Saarinen

Inside Linear: Building with taste, craft, and focus | Karri Saarinen (co-founder, designer, CEO)

Heuristic

A healthy researcher workload is two big projects plus one small side project; more than that gua...

A healthy researcher workload is two big projects plus one small side project; more than that guarantees mediocre work. Context: PM-researcher prioritization

Judd Antin

The UX Research reckoning is here | Judd Antin (Airbnb, Meta)

Frameworks from the people who've done it

Ranked by how many frameworks each practitioner contributes to this pack

Meng To 35 Jenny Wen 33 Karri Saarinen 27 Caitlin Kalinowski 25 Katie Dill 25 Dylan Field 24 Jessica Hische 23 Bob Baxley 20 Judd Antin 20 Ken Kocienda 15

Give your AI real expertise in design & craft

247 expert-sourced frameworks, rules, and principles. One .md file. Drop it in and your AI cites practitioners instead of guessing.

One-time purchase. No subscription. Instant download via email.