# Agents do the research. Every answer cites its source.

AI SaaS · Built solo, live at proux.design

**Every answer cites its source, across 350+ UX principles** · Live at proux.design

ProUX is an intelligent AI UX design platform. I built it solo, live at proux.design. You set a UX task and ProUX runs it. It picks the expert and the model, and names each choice on screen. Every claim links to its source, from 350+ UX principles. It measures a screen from its pixels instead of guessing.

## The problem

UX advice was opinion, so teams argued.

## What changed

- Grounded in verified guidelines, cited by number
- Audits measured from real pixels, scored out of 100
- 350+ UX principles, verified

## The result

Every answer cites its source, across 350+ UX principles · Live now

## Tab by tab

### The engine

Not a chatbot. You set a task and ProUX runs it.

- It reads 5 sources: experts with curated skills, your project context (goals, users and rules, read on every task), the guideline library (verified UX guidelines, retrieved by RAG), your screen (measured from pixels, not described) and the ProUX method (the steps every task runs through).
- It returns the right expert (chosen for the question, with its curated skills), the right model (picked for the job, yours to overrule) and informed design decisions, each one cited, with the reasoning behind it.
- Your product gets the same engine. Your rules, your data, decisions it can defend.

### It decides

You set the task. It picks the expert and the model.

- Every choice named on screen.
- One click changes it.
- The task: Audit this product page. Buyers drop before Add to cart. What do I fix first?
- Expert: Design audit, because you asked for an audit. Model: Auto · Sonnet 5, deep, picked by the job. Grounding: 3 project documents, your project first.
- Then it runs: analysing the design, searching ProUX guidelines, mapping findings. It used 16 guideline sources.

### Every source

Every claim links to its source.

- 1 tap opens the guideline and its evidence.
- When no source covers it, it says so.
- Darken the price text: it measures 2.43:1 against 4.5:1. Source 1: guideline 411, text legibility in eCommerce.
- Collapse long specs, so Add to cart stays in view. Source 2: guideline 32, collapse sections.
- Assumption: no project document says who buys.

### The audit

It measures the screen. It does not guess.

- Measured from the pixels, 43 elements read: price text 2.43:1, needs 4.5:1. Add to cart 36pt, needs 44pt. Body text 16px, passes.
- The score is 64, at risk: 100 less 24, 8 and 4. Critical 3, important 2, nice-to-have 2.
- Fixes pasted, measured again: price text fixed at 5.20:1, Add to cart fixed at 48pt. With 2 criticals fixed, the score moves from 64 to 80.

### ProUX vs general LLMs

Same task. A different kind of result.

- The same task, answered by a general LLM and by ProUX.
- A general LLM: “Looks light”, an adjective. ProUX: 2.43:1 against 4.5:1, measured from pixels.
- A general LLM: “Best practice”, no link. ProUX: guideline 411, 1 tap away.
- A general LLM: 8 equal tips, in no order. ProUX: ranked fixes, with the effort on each.
- A general LLM: a new answer each time. ProUX: the same rules on every page.
- A general LLM: start over in a new chat. ProUX: re-measured. Fixed, now 5.20:1.

[Try ProUX](https://proux.design)

Page: https://www.surinder.design/work/proux