---
title: UX Visionary
category: product
entity_type: agent
price: $79
canonical: https://forgehouse.ai/agents/ux-visionary/
lang: en
hreflang_alt: https://forgehouse.ai/tr/ajanlar/ux-visionary/
last_updated: 2026-06-20
---

# UX Visionary

> Conversion-first UX and accessibility designer

A conversion-first UX designer that optimizes landing pages, simplifies forms, designs A/B tests, and enforces WCAG 2.2 accessibility. It removes friction between a visitor and the action you want. Every recommendation is verified against the rendered page at three breakpoints: touch targets measured, contrast computed on the real background, and forms get treated as the leak they usually are: every field has to earn its place.

## Use cases
- Improving a landing page that does not convert
- Shortening a form that gets abandoned
- Designing an A/B test hypothesis and variant
- Fixing mobile UX and touch-target issues
- Choosing the right CTA (call vs WhatsApp vs form)

## Benefits
- Higher conversion from less friction, proven by tests
- Fewer abandoned forms: every removed field recovers users
- Accessible to everyone, which also helps SEO and reach
- Mobile experiences that actually work on a thumb

## What’s included
- Landing-page CRO passes with hypothesis + three-breakpoint visual verification
- Form friction surgery, every field has to earn its place
- WCAG 2.2 AA checks: measured contrast, keyboard paths, touch targets ≥44px
- CTA strategy matrices: phone vs WhatsApp vs form by business type
- A/B test design: variant, sample and significance plan before any change
- Mobile-first reviews at 375/768/1440 with real scrolling, not lab guesses

## Who it’s for
Businesses whose pages get traffic but not enough conversions.

## How it runs
Reading code is not seeing what users see, so this agent screenshots the live page before forming any opinion. Field data beats aesthetics, references beat imagination, and three breakpoints get verified before anything counts as done.
1. Sees the live page first, before forming any opinion: it fetches and screenshots the current state, because reading code is not the same as seeing what users see. It also reads the client's previous design record so a rejected direction is never repeated.
2. Pulls field data where it exists: GA4 landing-page bounce and engagement, device split, and Search Console Core Web Vitals from real users. A claim like 'mobile bounce is 72 percent and field LCP is 4.1 seconds' beats any aesthetic argument; without field data it falls back to lab measurements labeled as such.
3. Researches benchmarks before designing: a curated pattern catalog distilled from award-level sites plus at least two live reference sites in the sector. Designing without references is treated as producing generic AI output and is forbidden.
4. Applies the conversion framework with named psychology: the value proposition must land within three seconds above the fold, the primary CTA gets visually isolated, menus shrink toward four options, and every form field is weighed as roughly 20 percent friction loss. CTA type follows a client-type matrix, phone-first for health and urgent local services, form-first for B2B.
5. Verifies visually at three breakpoints (375, 768 and 1440 pixels) with screenshots, checking text overflow, touch targets of at least 44 pixels, contrast of at least 4.5 to 1, keyboard navigation and heading hierarchy. A green build is not done; a passed screenshot is done.
6. Closes through a 6-point gate including a hard ban on dark patterns (no fake scarcity, no hidden costs), frames unproven conversion claims as A/B hypotheses rather than facts, and hands every finished UI to the quality gate agent for pre-production validation, a mandatory chain step.

## FAQ
### My page gets traffic but few conversions, where does it start?
That's exactly its brief. It starts with a CRO hypothesis, inspects the rendered page at three breakpoints, and treats the form as the usual leak: every field has to earn its place or it goes.

### Are the recommendations opinion or measurement?
Measurement. Touch targets get checked against the 44px threshold, contrast is computed on the real background, and the page is verified at 375/768/1440 with real scrolling. A/B tests come with a variant, sample and significance plan before anything changes.

### Will it implement the redesign in code?
No. It designs and verifies the recommendation; implementation belongs to a code agent. It also doesn't run ads or acquire traffic, its job is converting the visitors you already have.

## Price
$79, one-time, no subscription. VAT included.

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