Persona Chat
AI-Powered User Persona Generator
Built an AI-powered SaaS platform that generates detailed user personas and enables designers to interact with them via natural conversation — accelerating research without real interviews.
The Problem
User research is time-consuming. UX designers often need to: • Create 4–6 detailed personas per project • Write dozens of interview questions • Simulate responses to validate assumptions This leads to: - Shallow research in early stages - Delayed product decisions - Gut-feel design instead of data-backed direction
The Solution
I built **Persona Chat** — a platform where designers could input a project description, and the system would: • Auto-generate multiple detailed personas (age, goals, behaviors, pain points) • Craft tailored questions for each persona • Simulate realistic, contextual responses using OpenAI • Enable real-time chat with any persona to explore product-market fit hypotheses This made early-stage research fast, scalable, and insightful.
My Role
Product
Collaborated with our design team to understand the UX research workflow. Defined what attributes mattered most in a persona and how to simulate 'believable' answers.
Backend
Built persona generation pipeline using OpenAI’s GPT-4 + prompt templates. Used MongoDB for persona/project storage and token-based access. Architected a conversation engine that retained chat memory per persona.
Frontend
Developed chat interface in React with personality-driven avatars and switching between multiple personas in the same project.
DevOps
Hosted on AWS with S3 static frontend, Lambda functions for OpenAI calls, and usage-based cost monitoring to manage API spend.
Key Features
AI-Generated Personas
Instantly generate multiple user types based on product input
Interactive Persona Chat
Simulate conversations with each persona to uncover insights
Automatic Question Builder
Smart generation of UX interview questions tailored to project goals
Realistic Conversational Tone
Each persona replies in a tone consistent with their traits and background
Project Workspace
Organize multiple personas, chat logs, and research materials in one place
Screenshots
Manage research projects, personas, and chat sessions
Real-time interaction with persona profiles via natural language
Outcome
🧪 **Used Internally by Designers**: Our team used the tool to validate early product ideas, cutting research prep time by 70%. 🧠 **High Concept Buy-In**: Designers loved the "talk to your user before you build" concept — strong feedback on realism and usability. 💸 **Paused for Strategic Pivot**: Development halted after company restructured priorities, but received interest from 3 external design firms for a white-labeled version.
Key Learnings
**Realism Matters in AI** — Personas must *feel* real, not just have accurate data. I learned to fine-tune tone, vocabulary, and contradictions to make conversations believable. **Cost Control is Critical** — Conversational AI can get expensive fast. We built usage caps, caching, and prompt cost estimates early on. **Design Tools Need Empathy** — UX professionals expect clarity and personality in their tools — tech alone doesn’t cut it.