№ 007Health & learningLive at testportrait.com
Testportrait — Voice Assessment Platform
A personality self-assessment you take by talking, scored by AI into a personal report.
An AI-assisted voice assessment platform exploring personality patterns and cognitive style. The existing case study documents the assessment workflow, technical architecture and reported project results.
- Voice AI
- Speech processing
- AI assessment
- Web applications
The problem
Questionnaires are tedious and easy to game. A spoken conversation is more natural and gives richer material to work with.
Who it is for
People curious about their own patterns. A paid consumer product.
What it does
01
Talk, don't type
A voice agent runs the conversation.
02
The report
Claude scores the transcript against a structured framework and writes a personalised PDF.
03
Accounts and history
Every report stays in the account.
04
Payments and a book
Stripe unlock or subscription, and a companion book store.
How it works
A Next.js app on Vercel. ElevenLabs runs the voice agent, Claude scores the transcript, Supabase holds accounts and reports, Stripe takes payment, react-pdf renders the report. An admin back office sits behind it.
- Next.js 15
- ElevenLabs
- Claude
- Supabase
- Stripe
- react-pdf
- Role
- Built by Vas.
- Status
- Live at testportrait.com.
The Challenge
The problem
The client wanted to create an engaging psychological assessment that felt natural and modern. Traditional text-based tests felt impersonal and required too much effort from users.
Why existing solutions failed
Text-based personality tests require reading and typing, which feels like work. The experience is clinical rather than conversational, making it less engaging and harder to complete.
Business impact
The goal was to create something people would actually want to do and share with friends - a personality test that feels more like a fun conversation than a questionnaire.
The workflow
From first interaction to final output, step by step.
- STEP 01
User Opens App
Start conversation immediately with no signup required
- STEP 02
AI Asks Questions
Natural voice conversation about imaginary scenarios
- STEP 03
User Responds
Speak naturally while AI listens and processes
- STEP 04
Real-time Visualizations
See voice activity with engaging visual feedback
- STEP 05
AI Analysis
Claude AI analyzes responses for personality insights
- STEP 06
Generate Report
Create comprehensive personality assessment
- STEP 07
View Results
Get instant insights about your mind's architecture
Our Solution
Approach
Created a voice-first platform where users simply talk to an AI. The AI asks questions about imaginary scenarios (cubes, horses, deserts), listens to responses, and analyzes the conversation to reveal personality insights. Users can start anonymously and get their results immediately.
Workflow
User opens app → Start conversation immediately → AI asks questions → User responds with voice → Real-time voice visualizations → AI analyzes responses → Generate personality report → View results instantly
Key features
- Natural voice conversation with AI
- No typing or forms required
- Start anonymously - no signup needed
- Real-time voice visualizations
- Instant AI-powered personality insights
- Beautiful, shareable results
- Optional account to save results
- Works on desktop and mobile
Results
Before
No existing solution - this was a new product
After
Launched Testportrait with 10,000+ assessments completed. Users get professional personality insights through a 5-minute voice conversation.
What changed for the business
Successfully launched a voice-first personality assessment platform that has completed over 10,000 assessments. The 5-minute conversation format makes it accessible and engaging, while AI-powered analysis delivers professional-grade insights instantly.
Technical Details
Architecture
React SPA with TypeScript and Vite for the frontend. ElevenLabs Conversational AI SDK handles real-time voice interactions. Supabase provides database, authentication, and magic link delivery. Webhook receives conversation transcripts and triggers Claude AI analysis via Anthropic API. WebAudio API creates real-time voice visualizations. Anonymous users get UUID-based sessions with optional account creation.
Integrations
- ElevenLabs Conversational AI SDK for voice interactions
- Claude AI (Anthropic) for transcript analysis
- Supabase for database and authentication
- Magic link authentication system
- WebAudio API for real-time visualizations
- Webhook endpoint for async processing
- UUID-based anonymous session management
Security and compliance
- Anti-enumeration magic links
- Rate limiting on all endpoints
- Audit logs for all actions
- Secure webhook authentication
- Anonymous UUID sessions for privacy
- HTTPS/TLS for all communications
- Secure storage of conversation data
- Optional account creation for data persistence
Technologies
- React
- TypeScript
- Vite
- ElevenLabs Conversational AI SDK
- Claude AI (Anthropic)
- Supabase
- WebAudio API
- Magic Link Authentication
- Webhooks
- Tailwind CSS
Related
№ 008In production at a clinicMedical Clinic Fax Automation
Document processing for medical faxes, including extraction, classification and human verification. Related work includes the clinic application, fax dashboard prototypes, cloud infrastructure and eClinicalWorks integration documentation.
- OCR
- AI extraction
- AWS
- Workflow automation
№ 003Live in productionPFC Coaching Platform
Coach and client workspaces for training programs, nutrition, progress tracking and messaging. Includes client onboarding and check-in automation, a mobile web experience and role-based access.
- Next.js
- React
- Convex
- Clerk
- PWA
Next project
Personal OSA solo business owner's workspace with an assistant that remembers everything and drafts the admin.