A platform to validate designs before vibe-coding

A platform to validate designs before vibe-coding

Anyone can ship a product with AI. Almost no one knows if it works for their audience.

Full Product

The Gap…

The Gap…

The AI builder wave changed who can ship. Lovable, v0, Bolt, Cursor — a solo founder can go from idea to live product in days. But speed created a new blind spot.

These products reach real users without ever being tested against the actual humans they're meant for. Traditional user testing assumes you have weeks, a budget, and a research team. Early-stage founders have none of those.

So most just guess. And find out they were wrong after launch.

The question no tool was answering: Does this design actually work for my specific industry and audience — before I spend money acquiring users?

The Gap…

The AI builder wave changed who can ship. Lovable, v0, Bolt, Cursor — a solo founder can go from idea to live product in days. But speed created a new blind spot.

These products reach real users without ever being tested against the actual humans they're meant for. Traditional user testing assumes you have weeks, a budget, and a research team. Early-stage founders have none of those.

So most just guess. And find out they were wrong after launch.

The question no tool was answering: Does this design actually work for my specific industry and audience — before I spend money acquiring users?

Problem

The AI builder wave changed who can ship. Lovable, v0, Bolt, Cursor — a solo founder can go from idea to live product in days. But speed created a new blind spot.

These products reach real users without ever being tested against the actual humans they're meant for. Traditional user testing assumes you have weeks, a budget, and a research team. Early-stage founders have none of those.

So most just guess. And find out they were wrong after launch.

The question no tool was answering: Does this design actually work for my specific industry and audience — before I spend money acquiring users?

The AI builder wave changed who can ship. Lovable, v0, Bolt, Cursor — a solo founder can go from idea to live product in days. But speed created a new blind spot.

These products reach real users without ever being tested against the actual humans they're meant for. Traditional user testing assumes you have weeks, a budget, and a research team. Early-stage founders have none of those.

So most just guess. And find out they were wrong after launch.

The question no tool was answering: Does this design actually work for my specific industry and audience — before I spend money acquiring users?

Insight

The problem isn't just slow feedback. It's that feedback was never industry-aware.

A checkout flow for a B2C fashion app needs to feel different than one for a B2B SaaS tool. A 22-year-old mobile-first shopper has completely different expectations than a 45-year-old first-time user. Generic usability testing misses this — it tells you what's broken but not why it's wrong for your market.

That gap is where PersonaLab lives.

Solution

PersonaLab generates AI personas calibrated to your exact industry and target audience, simulates how they interact with your design, and returns a prioritized issues breakdown — in under 30 seconds.

Input: Paste a Figma prototype link or live URL → define your industry, business model, and target audience.

Process: AI generates 3–5 personas that match your market context. Each one simulates interaction with your design and surfaces friction points based on who they are.

Output: A dashboard showing persona-by-persona feedback, an issues breakdown ranked by severity and business impact, and a UX score. Every session is saved to version history so you can fix, re-test, and track improvement.

Role

AI Product Designer - I worked on PRD, to visual designs to system design flows to code a workable web app via AI tool called Lovable.

AI Product Designer - I worked on PRD, to visual designs to system design flows to code a workable web app via AI tool called Lovable.

Tools used

ChatGPT, Claude, Lovable, Github, Figma, and FigJam

ChatGPT, Claude, Lovable, Github, Figma, and FigJam

Duration

1-week concept sprint

1-week concept sprint

Not for UX researchers. They already have tools. But for

Not for UX researchers. They already have tools. But for

Not for UX researchers. They already have tools. But for

1

The founder who built their MVP with Lovable and needs a gut-check before launch

2

The indie developer who shipped a landing page with v0 and isn't sure if it converts

3

The early-stage startup with real users showing up tomorrow and zero research budget

4

The solo designer who needs faster iteration feedback without recruiting participants every cycle

Key Product Decisions

Key Product Decisions

Industry + audience context comes first Without it, AI feedback is generic. Forcing users to define their market before running a test means the personas generated are calibrated to their actual product context — not a fictional average user. This is the feature that makes the output feel relevant instead of generic.

Named personas with distinct archetypes Sarah (tech-savvy, power user), Mike (beginner, new user), Emma (mobile-first user) represent the axes of variation that matter most for early products: expertise level and device context. Separating feedback by persona makes it immediately visible which user segment is most at risk — and why.

"What to Fix?" leads the dashboard Early founders are overwhelmed. The overview screen leads with one critical issue and a clear CTA — not a wall of data. The insight from testing: people don't want all the information, they want to know what to do next.

Industry + audience context comes first - Without it, AI feedback is generic. Forcing users to define their market before running a test means the personas generated are calibrated to their actual product context — not a fictional average user. This is the feature that makes the output feel relevant instead of generic.

Named personas with distinct archetypes- Sarah (tech-savvy, power user), Mike (beginner, new user), Emma (mobile-first user) represent the axes of variation that matter most for early products: expertise level and device context. Separating feedback by persona makes it immediately visible which user segment is most at risk — and why.

"What to Fix?" leads the dashboard Early founders are overwhelmed. The overview screen leads with one critical issue and a clear CTA — not a wall of data. The insight from testing: people don't want all the information, they want to know what to do next.

Strategy & Ideation

Strategy & Ideation

• Developed the core product concept and problem validation
• Created comprehensive PRD with user stories and technical requirements

• Defined MVP features and future roadmap

• Developed the core product concept and problem validation
• Created comprehensive PRD with user stories and technical requirements

• Defined MVP features and future roadmap

• Developed the core product concept and problem validation
• Created comprehensive PRD with user stories and technical requirements

• Defined MVP features and future roadmap

See Full PRD

System Design Diagram

System Design
Diagram

• Flow overview : Upload Figma prototype → AI personas interact → Get feedback in under 30 seconds
• Focus on generating 3-5 diverse user types that behave like real people from different demographics and industries

• Goals was scalability as well- to support both solo designers and enterprise at the same performance level

• Flow overview : Upload Figma prototype → AI personas interact → Get feedback in under 30 seconds
• Focus on generating 3-5 diverse user types that behave like real people from different demographics and industries

• Goals was scalability as well- to support both solo designers and enterprise at the same performance level

• Flow overview : Upload Figma prototype → AI personas interact → Get feedback in under 30 seconds
• Focus on generating 3-5 diverse user types that behave like real people from different demographics and industries

• Goals was scalability as well- to support both solo designers and enterprise at the same performance level

See Complete Flow

Lessons Learned

Lessons Learned

Lessons
Learned

  1. Users don’t want all the data—they want to know what to do next.


  2. It’s better to ship an imperfect solution, learn, and refine than to over-polish in isolation- I wasted a lot of time, effort and money in building fast than refining PRD


  3. Using AI to simulate user personas taught me how AI can actively shape UX research and reduce dependencies on traditional, costly processes.

  1. Users don’t want all the data—they want to know what to do next.


  2. It’s better to ship an imperfect solution, learn, and refine than to over-polish in isolation- I wasted a lot of time, effort and money in building fast than refining PRD


  3. Using AI to simulate user personas taught me how AI can actively shape UX research and reduce dependencies on traditional, costly processes.

  1. Users don’t want all the data—they want to know what to do next.


  2. It’s better to ship an imperfect solution, learn, and refine than to over-polish in isolation- I wasted a lot of time, effort and money in building fast than refining PRD


  3. Using AI to simulate user personas taught me how AI can actively shape UX research and reduce dependencies on traditional, costly processes.

Next Case Study ->

Create a free website with Framer, the website builder loved by startups, designers and agencies.