Caroline Xia
PUBLIC CASE STUDY

Petsona: A Social App for Your Pet's AI Persona

Give every pet an AI persona—and let compatible companions find each other nearby.

2026

Project Snapshot

Date
2026.02-2026.06
Role
Co-founder, Full-stack Developer
Status
Public case study
Award
3rd Prize, SG TRAE Friend Hackathon

Overview

Petsona explores pet social networking through an identity that belongs to the pet rather than only its owner. The AI persona expresses the pet's habits and temperament while remaining grounded in information provided by the owner.

My Role

As co-founder and full-stack developer, I shaped the product strategy and implemented the journey from persona creation and candidate discovery to matching and conversation.

Problem

Pet owners want safe, compatible companions nearby but cannot easily judge whether two animals will get along. Starting a conversation with an unfamiliar owner adds another layer of friction.

Solution

Photos, a personality questionnaire, and owner descriptions generate a pet persona with an expression style, preferences, and boundaries. Nearby discovery applies rule-based safety filters before personality-aware agents explain compatibility and generate an opening message.

Core Flow

Upload a photo and complete the questionnaire → create the persona and long-term memory → request nearby matching → filter candidates by distance, type, and safety state → pre-rank across five compatibility dimensions → let two pet agents evaluate the match → present a recommendation card and personalized icebreaker → update pet or relationship memory after conversation.

Outcome

Petsona won third prize at the SG TRAE Friend Hackathon.

Media & Resources

Petsona persona creation flow
From pet inputs to a stable, expressive AI persona.
Petsona personality card
A shareable representation of personality, habits, and preferences.
Petsona nearby matching experience
Nearby candidates, compatibility reasoning, and an AI-generated icebreaker.
Petsona agent architecture
The persona, memory, matching, and conversation architecture.