---
title: Refero Design
category: product
entity_type: skill
price: $15
canonical: https://forgehouse.ai/skills/refero-design/
lang: en
hreflang_alt: https://forgehouse.ai/tr/skiller/refero-design/
last_updated: 2026-06-20
---

# Refero Design

> Research-First design methodology using Refero MCP.

A research-first design methodology that grounds every interface decision in real product screens instead of guesswork. It runs a structured four-phase workflow: Discover, Research, Analyze, Design, Implement, so your AI builds screens with the craft of a senior product designer, not the generic 'AI slop' that every model defaults to. The result is UI with a rock-solid foundation plus the distinctive soul that makes it memorable.

## Use cases
- Designing a new pricing, onboarding, or landing screen from scratch
- Escaping generic AI-default designs (indigo gradients, blob backgrounds, perfect symmetry)
- Researching how top products solve a specific UI problem before building
- Extracting exact patterns: real copy, numbers, conditions: from leading apps
- Building a persuasion layer: hook, objection killers, trust signals before any code
- Running a side-by-side quality gate against best-in-class references

## Benefits
- Ships interfaces that look intentionally crafted, not machine-generated
- Replaces blind guessing with a defensible design brief and documented reasoning
- Forces a Steal List of 5+ concrete tactics so designs convert, not just decorate
- Builds typography, color, spacing, and motion to senior-level discipline every time

## What’s included
- 8-question discovery framework that outputs a one-line design brief
- Search strategy with eight query types (broad, leader, component, adjacent, emotion)
- Three analysis lenses: structure, visual craft, conversion and soul
- Persuasion-layer table: hook, story arc, objection killers, trust signals
- Anti-AI-slop checklist with explicit do/don't rules (no default indigo)
- Quality gate plus side-by-side reference comparison for final validation

## Who it’s for
Designers, founders, and builders who want AI-generated UI that reads as professionally crafted and on-brand rather than templated.

## How it runs
Before a pixel moves, 50 to 100 real product screens get studied. Research becomes a steal list of exact tactics, design follows hard craft rules, and the build is validated against the references.
1. Opens with a discovery brief before any design: what is being built, for whom, the job the page is hired to do, the main objection to overcome and the one thing users should remember tomorrow.
2. Researches the real landscape through the Refero corpus: 5+ query angles (broad, element, company, style, platform), 50-100 screens scanned, then deep get_screen analysis on the 5-10 best finds, because the gold hides in the long tail past the first 10 results.
3. Synthesizes findings into a pattern table and a Steal List of at least 5 tactics with exact copy, numbers and conditions. Generic findings like clean design are banned; 13px/20px body with -0.01em tracking passes.
4. Designs with hard craft rules: a ratio-based type scale with mandatory letter-spacing per text type, a 4-layer color palette with default indigo explicitly forbidden, 4px/8px spacing rhythm and motion capped at 90-360ms by category.
5. Fills the persuasion layer before writing code: the 3-second hook, objection killers, minimum 2 trust signals and the one memorable element, then adds the 20 percent soul on top of the 80 percent proven patterns.
6. Validates the build side by side against the top 3 references and a quality gate covering function, visual craft, persuasion and polish; if something feels off, it loops back to research instead of guessing.

## FAQ
### Can I use this without the Refero MCP connected?
Partly. The discovery questions, analysis lenses, and quality gate read as a standalone methodology, but the Research phase pulls real product screens through Refero. Without it, you would have to feed that phase with reference material you collect yourself.

### How does it actually prevent generic 'AI slop' designs?
Three mechanisms: an anti-slop checklist with explicit do/don't rules (no default indigo, no blob backgrounds), a mandatory Steal List of five-plus concrete tactics extracted from real apps, and a side-by-side quality gate against best-in-class references. Decisions trace to documented reasoning, not model defaults.

### Does it output finished visual mockups or Figma files?
No. Its outputs are a design brief, pattern research, and a persuasion layer: hook, objection killers, trust signals, that your build step then implements. It is a methodology, not an image generator.

## Price
$15, one-time, no subscription. VAT included.

Related guide: [Building a design system with Claude Design](https://forgehouse.ai/guides/claude-design-design-system/)