Personity

Inspiration

As a Business Administration student specializing in Data Analytics, I was taught that surveys provide speed but interviews capture depth. To truly understand users, one needs both. The idea stayed with me until one day, while using ChatGPT, the thought emerged: What if AI could conduct user interviews, combining both speed and depth? That question led to the creation of Personity.

What it does

Personity is an AI-powered conversational research platform that transforms traditional static surveys into adaptive, interview-style conversations.

Instead of fixed questions, Personity:

Asks intelligent follow-up questions, Probes vague or unclear responses, Detects contradictions, Adapts question flow in real time, Extracts quotes, personas, themes, and sentiment, Generates an automatic research report. Users can create a research survey in under five minutes, share a link, and respondents engage in a conversational experience that they can pause and resume. This leads to significantly higher completion rates than traditional surveys. The outcome is the depth of qualitative interviews and the scale of quantitative surveys, achieved through AI.

How we built it

I do not know how to code, but using Kiro, I was able to build a production-grade AI SaaS product.

The process included: Drafting specifications and requirements using AI, Designing the UI with Aceternity and Shadcn UI components, Integrating Supabase MCP for backend functionality, Adding Razorpay for payments and PostHog for analytics, Deploying on Vercel, Building a guided wizard to create surveys quickly.

Challenges we ran into

One of the most complex challenges was designing the AI master prompt. It took 11 versions of iteration to make the AI behave like a real researcher who can track context, adapt questions, and summarize insights effectively.

Other challenges included: Building without prior coding experience, Understanding Supabase MCP and API integrations, Designing a structured prompt system for AI reasoning, Optimizing token usage for cost efficiency, Creating a SaaS-level UX and workflow.

Accomplishments that we're proud of

50+ working SaaS features, Dark and light mode UI, Guided research wizard, Insights dashboard with personas, themes, quotes, and sentiment, Automatic PDF and CSV export, Fully functional payments and analytics, Production-grade prompt system for AI-led research.

What we learned

AI can replicate structured research workflows, Prompt engineering can simulate real researcher behavior, Good UX builds trust as much as AI performance does, Building without code is possible with the light AI tools, Structured context handling is critical for scalable research.

What's next for Personity

The next phase is to enable WhatsApp-based research for India, allowing respondents to interact through:

Typed chat messages, Voice responses in regional languages, Conversational flows using familiar communication platforms, This will make AI-led research accessible across dialects, regions, and literacy levels. The long-term vision is to create the new standard for research, depth at scale.

Built With

• apiroutes
• autoprefixer
• axios
• azureopenai
• azureopenaiapi
• babelreactcompiler
• bcryptjs
• caching
• classvarianceauthority
• clientcomponents
• clsx
• codesplitting
• css
• dotenv
• edgefunctions
• eslint
• fontsource/inter
• framermotion
• google-auth-library
• googlegemini
• googlegeminiapi
• googleoauth
• googleoauthapi
• gpt-4o
• imageoptimization
• inputvalidation
• javascript
• jsonwebtoken
• jspdf
• jwt
• kiroai
• lazyloading
• lucidereact
• mammoth
• mcp
• middleware
• nanoid
• next.js
• node.js
• openaisdk
• pdf-parse
• postgresql
• posthog
• posthogapi
• prettier
• radixui
• ratelimiting
• razorpay
• razorpayapi
• react
• reacthookform
• recharts
• resend
• resendapi
• routegroups
• servercomponents
• shadcn/ui
• sql
• supabase
• supabaseapi
• supabaseauth
• supabasestorage
• tailwindcss
• tailwindmerge
• tanstackquery
• tsx
• typescript
• upstash
• upstash/ratelimit
• upstashredis
• vercel
• zod
• zustand