How This Gen Z Founder Raised $11 Million To Build The AI-Powered Closet From 'Clueless'

Key Concepts

Ulta: An AI stylist application that provides daily outfit recommendations based on a user's closet, personal style, calendar, and weather.
AI Stylist: An artificial intelligence system designed to offer fashion advice and outfit suggestions.
Transformer Models & Diffusion Models: Advanced AI architectures that are crucial for Ulta's capabilities in understanding and generating fashion-related content.
Multimodal AI: AI systems that can process and integrate information from various sources, such as text, images, and video, which is essential for fashion.
Consumer AI: Artificial intelligence products designed for end-users, focusing on utility, entertainment, and user satisfaction.
Democratizing Technology: The concept of making advanced technologies, like personal styling, accessible to a wider audience through AI.
Shoppertainment: The fusion of shopping with entertainment, where fashion is presented as a fun form of self-expression.
Cost Per Wear: A metric that helps users understand the value of their clothing items by dividing the purchase price by the number of times the item has been worn.
Multiplayer AI: AI experiences that involve multiple users interacting with the AI and each other, a frontier Ulta aims to explore.

Ulta: The AI Stylist Revolutionizing Fashion

This summary details the creation and functionality of Ulta, an AI-powered fashion application, as explained by its founder. The company leverages cutting-edge AI technologies, specifically transformer and diffusion models, to address the personal challenge of deciding what to wear daily.

The Genesis of Ulta: A Personal Problem Solved by AI

The idea for Ulta stemmed from the founder's personal struggle with daily outfit selection, a common issue characterized by piles of clothes and complex decision-making processes involving outfit repetition, weather, and social context. The inspiration was further fueled by the iconic "Clueless" movie, which envisioned a futuristic closet system. The founder recognized that advancements in AI, particularly transformer and diffusion models, had finally reached a point where such a product could be realized. This led to the establishment of Ulta, aiming to be both a utilitarian AI product and an enjoyable fashion companion.

User Experience and Technological Underpinnings

Consumer-Facing Features:

Closet Digitization: Users can upload photos of their clothing from their camera roll or social media. The AI can recognize garments and replace them with original product images for a clean catalog. Email receipts can also be forwarded to automatically populate the closet.
Automated Tagging: Ulta automatically handles background removal, tagging of fabrics, dress codes, colors, and other attributes for each clothing item.
Personalized Styling: Users can train their AI stylist through natural language feedback, specifying preferences like monochrome looks on Mondays or avoiding certain color combinations.
Avatar Generation: Users can upload photos of themselves to create a personalized avatar, on which Ulta generates outfits, providing a visual representation of how the clothes would look on them.
Prompt-Based Styling: The app allows users to request outfits for various occasions through natural language prompts, ranging from casual outings to formal events.
Additional Features: Ulta can also generate packing lists and lookbooks.

Backend Technology:

Complex AI Architecture: Ulta employs nearly a dozen AI models to handle the intricate taxonomy of fashion. This includes understanding visual attributes like the placement of embellishments and fabric textures.
Color Nuance Model: Recognizing that human color perception is vast (around a million colors), Ulta has an in-house model that assesses the probability of specific shades of colors complementing each other on a particular individual.
Item and Style Understanding: The AI models are trained to understand individual clothing items, their suitability for different occasions, and how to pair them. Crucially, they aim to understand a user's personal style.
Focus on User Experience: The founder emphasizes that for consumers, the underlying AI models are less important than the product's utility, fun factor, and ability to make them happy.

The Founder's Journey: From Fashion Enthusiast to Tech Innovator

The founder's passion for fashion began in her youth, with early engagement on platforms like Polyvore and Pinterest, and a habit of collecting fashion magazine cutouts. Despite a background in computer science and math, she initially didn't see a direct path into the fashion industry. A pivotal moment was winning the Gucci hackathon, which provided her first industry contact and mentorship. This experience opened doors to networking with fashion professionals, leading to a realization of the significant overlap between fashion and technology. Many of these early contacts are now investors in Ulta.

Fundraising and Investment Appeal

Ulta has successfully raised $11 million from a diverse range of investors, including tech-focused firms like Menlo Ventures and fashion conglomerates like the LVMH family. The core pitch to these investors centered on:

Massive Market Opportunity: Fashion, e-commerce, and retail represent a substantial industry, with plans to expand into beauty and other sectors.
Evolving Consumer Behavior: The shift from department stores to e-commerce, mobile, and now AI applications necessitates a team capable of navigating these changes.
Dual Expertise: The team possesses both a deep understanding of fashion and the technical prowess to execute complex AI solutions.
Democratizing Personal Styling: The vision is to use AI to provide a concierge-level styling experience to all shoppers, not just the wealthy or celebrities.

The Impact of Advanced AI Models

The "T" in GPT (Generative Pre-trained Transformer) has been instrumental in Ulta's development. However, the founder notes that Large Language Models (LLMs) alone are insufficient for fashion due to its multimodal nature. Fashion involves understanding imagery and video to capture how garments fit, flow, and look in motion. Each advancement in AI models has enabled Ulta to unlock new features, such as the highly popular avatar generation, which can now produce an avatar every 1-1.5 seconds at peak times. Earlier iterations of diffusion models struggled with rendering specific brand logos accurately (e.g., "gucci" appearing as "gucuci"), but current models can render intricate details like the "LVMH" logo on a Gucci bag.

Revenue Model and Future Expansion

Ulta's current revenue model is affiliate-based. When a user purchases a recommended item through the app, Ulta earns a commission. The founder sees significant potential for future revenue streams, including:

Product Placement: Strategic integration of brands and products.
Personalized Branded Content: Creating tailored content for brands.
Premium Features: Offering access to more advanced models or exclusive features for a fee.

Gen Z's Influence on Fashion and Shopping Habits

Gen Z users have become a significant demographic for Ulta, interacting with the app differently than older users. This AI-native generation expects AI-driven shopping experiences. Their requests include features like:

Avatar Customization: Dressing avatars with school logos for events like tailgates and game days.
Integrated Experiences: Combining music with outfit generation and shopping.

This reflects Gen Z's unique shopping behaviors and desires, which Ulta is actively incorporating into its development.

Sustainability and Reinvigorating Closets

The statistic that users typically wear only about 20% of their closet highlights the issue of fashion waste. Gen Z's growing concern for sustainability makes Ulta's ability to help users maximize their existing wardrobes particularly appealing. The "cost per wear" feature, which calculates the value of an item based on its usage, is popular. Furthermore, Ulta is partnering with Poshmark to offer suggestions for secondhand shopping within the app, promoting a more sustainable approach to fashion.

The Future of AI in Fashion: Personalization and Shoppertainment

The intersection of AI, tech, and fashion is expected to lead to increasingly personalized shopping experiences, offering a concierge-like service to all. Ulta aims to blend utility with entertainment, categorizing this as "shopper-tainment." The founder highlights the profound impact fashion has on self-confidence and how users report Ulta helping them regain their sense of self, especially during challenging times like long COVID recovery.

Advice for Young Female Founders

The founder offers advice for aspiring female founders:

Continuous Learning: Actively seek knowledge from technical advisors, investors, and the user community.
Team Building: Cultivate a team that is intelligent, supportive, and feels like family.

Exciting Future Developments for Ulta

Looking ahead, Ulta is focused on:

Social Features: Pioneering "multiplayer AI" where users can interact with the AI and each other.
Expansion into Beauty: Addressing user requests for hair, makeup, and fragrance recommendations to complement outfits.
Evolving Retail Landscape: Adapting to changes in retail and consumer purchasing behaviors.

The Role of Youth in Entrepreneurship

The founder views being under 30 as a significant advantage, allowing for intense focus on building the company without the obligations of children or pets. This enables the dedication required for the demanding startup culture. Despite the intensity, the positive feedback from users worldwide fuels her motivation and prevents burnout.