HeatChat
Designing an AI companion for Heat
Role
Concept, UX Design, UX Writing, AI Training, Brand Voice
Overview
HeatChat is the AI assistant powering Heat’s upcoming app. I created its voice, personality, meta-human identity, and led the branding and launch strategy to make it feel alive and connected to the product.
Deliverables
HeatChat AI concept and persona
UX design and interface for chat interactions
Prompt library and AI training examples
Integrated chat within the Heat app, including Missed Connections and Horoscope
Voice guide and copy framework for consistent communication
Marketing and launch strategy: social media campaigns, community engagement, and a launch event to introduce HeatChat to users
Approach
I led the design and implementation of HeatChat end-to-end:
Concept & Voice: Defined a persona blending class and street awareness, giving the AI a personality that is knowledgeable, approachable, and playful.
AI Training: Wrote prompt templates and curated training examples to reinforce tone and behavior.
UX Design: Designed the chat interface, message flow, and integration points in the app.
Community Feedback: Asked followers how they pictured the AI (male/female, tone, personality) and incorporated their input.
Iteration: Tested messages and edge cases to ensure helpfulness, consistency, and entertainment value.
Problem
HeatChat needed to feel alive and engaging.
Standard AI tools sounded robotic and generic.
Users needed a voice they could trust for city recommendations and social insights.
The AI had to balance personality, knowledge, and clarity.
Process
Research and Community Feedback →
I knew assumptions alone were not enough. HeatChat needed to reflect real user expectations, not just internal ideas of what an AI assistant should be.
To ground the concept, I asked our community how they pictured the AI. Questions focused on personality and presence, such as whether it felt more masculine or feminine, sophisticated or playful, and classy or street smart. These responses revealed a clear desire for something relatable, confident, and human.
Using this input, I defined a persona that blended class with street awareness. The voice was designed to feel informed but not condescending, playful without being gimmicky, and confident without sounding scripted.
This personality guided every decision throughout the experience, from how the AI greeted users to how it handled humor, recommendations, and moments of uncertainty. The result was an assistant that felt more like a trusted local than a chatbot.
AI Training
To bring the personality to life, I wrote structured prompt templates and curated training examples that reinforced tone, pacing, and behavioral patterns. These prompts were designed to adapt across different contexts while maintaining a consistent voice.
I focused on teaching the AI how to respond, not just what to say. This included how to handle ambiguity, how to recover from misunderstandings, and how to maintain warmth even when delivering practical information.
UX Design
I designed the chat interface, message flow, and key integration points to feel lightweight and natural within the app. The experience needed to feel like an extension of the product, not a separate tool.
Special attention was given to message timing, response length, and visual hierarchy to avoid overwhelming users. The goal was to create a conversational rhythm that felt effortless and encouraged continued interaction.
Iteration and Testing
I continuously tested messages, edge cases, and conversational paths to ensure the AI remained helpful, consistent, and engaging. This included stress testing unusual requests, tone shifts, and long conversations.
Each iteration refined clarity, reduced friction, and strengthened personality. The final experience balanced usefulness with entertainment, encouraging users to return not just for answers, but for the interaction itself.
Outcome
Users felt HeatChat was approachable and fun while remaining informative
New functionalities increased engagement and delight:
Missed Connections helped users reconnect in meaningful ways
Horoscope added personalized, playful recommendations
Early tests and community feedback confirmed the persona and tone were successful
HeatChat became a core part of the Heat experience, shaping how users interact with the app’s AI