hims & hers SUPPLEMENTAL PRESENTATION Investor Presentation August 2025 > INTRODUCTION Safe Harbor Statement This presentation has been prepared by Hims & Hers Health, Inc. ("we," "us," "us," "Hims & Hers," or the "Company"). This presentation includes forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended and performance to be materially different from those expressed or implied by these forward-looking statements. These risks and Section 21E of the Securitie ...

Industry Trend - AI is poised to democratize access to personalized wellness services, making them available to a broader population at a significantly lower cost [1] - AI and AI-powered products are expected to substantially reduce the cost of wellness-related offerings, particularly those delivered digitally [2] Wellness Service - Currently, luxury wellness includes services like wellness coaches and personal trainers, which are expensive and accessible only to a small fraction of the population [1]

AI Personalization & User Experience - AI predicts user preferences on platforms like Netflix and TikTok, shaping user experience [1] - AI personalization creates "filter bubbles," limiting exposure to diverse content and perspectives [4][6] - Algorithms on platforms like TikTok can quickly polarize users with political content, creating echo chambers [6] - Over-personalization can confine users to comfort zones, hindering discovery of new interests [5] Impact on Personal Growth & Learning - AI-driven education apps, while helpful, may prevent users from confronting real challenges and limit progress [10] - Social media algorithms trap users, influencing attention and potentially leading to excessive usage [11] Recommendations & Mitigation Strategies - Users should actively explore diverse content on platforms like Spotify and Netflix to break out of filter bubbles [12] - Disabling autoplay features on platforms like YouTube can help users regain control over their choices [13] - Users can reset or tweak algorithms on platforms like Spotify to diversify recommendations [13] - The industry should promote responsible AI usage to expand user horizons and encourage exploration [14]

Foundation Model Strategy - Netflix is leveraging foundation models for personalized recommendations [1] - The strategy is based on work by Yesu Feng, a staff research scientist/engineer at Netflix, focused on generative foundation models [1] - Prior to Netflix, Feng worked on feed and marketplace optimization at LinkedIn and Uber, respectively [1] Industry Focus - The application of foundation models aims to improve personalized recommendations [1] - The discussion took place at the AI Engineer World's Fair in San Francisco [1]

Model Building and Training - LinkedIn leverages large language models (LLMs) for personalization and ranking tasks, aiming to use one model for all tasks [2][3] - The process involves converting user information into prompts, a method called "promptification" [8] - LinkedIn builds a large foundation model, Blue XL, with 150 billion parameters, then distills it to smaller, more efficient models like a 3B model for production [12] - Distillation from a large model is more effective than training a small model from scratch [14] - Increasing data, model size (up to 8x22B), and context length can improve model performance, but longer contexts may require model adjustments [17][18][19] Model Performance and Generalization - The model improves performance for cold start users, showing a growing gap compared to production models as interactions decrease [21] - The model demonstrates generalization to new domains, performing on par with or better than task-specific production models in out-of-domain tasks [23] Model Serving and Optimization - LinkedIn focuses on model specification, pruning, and quantization to improve throughput and reduce latency for production [26] - Gradual pruning and distillation are more effective than aggressive pruning, minimizing information loss [29][30] - Mixed precision, including FP8 for activations and model parameters but FP32 for the LM head, is crucial for maintaining prediction precision [31][32] - Sparsifying attention scores can reduce latency by allowing multiple item recommendations without each item attending to each other [34][35] - LinkedIn achieved a 7x reduction in latency and a 30x increase in throughput per GPU through these optimization techniques [36]

AI in Retail & Fashion: Key Takeaways - AI is transforming retail and fashion across three horizons: next year, three years from now, and 2030 [1] - Retailers are currently spending 1-5% of their revenues on customer service, presenting a significant opportunity for AI implementation [1] - AI can enhance customer service by processing customer policies, product recommendations, and interacting via phone, email, and chat [1] - AI agents will assist human agents in handling sensitive issues or approvals [1] Personalized Customer Experience - By 2028, AI will enable digital twins, using clean data and data links to create hyper-personalized offers and product recommendations [1] - By 2030, AI will proactively suggest purchases through wearable devices and smart homes, creating fluid customer journeys [1][3] - AI will not only identify customer needs but also upsell them to the next level [3] Impact on Employees - AI empowers employees like personal shoppers with real-time insights into customer preferences and sales [2][4] - AI will be the foundation for both customer and employee experiences in the next five years [4]

Core Insights - Stitch Fix, Inc. (SFIX) has achieved a 10% year-over-year growth in Average Order Value (AOV) in the third quarter of fiscal 2025, marking the seventh consecutive quarter of growth [1][9] Group 1: AOV Growth Drivers - The introduction of larger Fixes, allowing clients to receive up to eight items, has significantly contributed to AOV growth, with adoption more than doubling since the first quarter [2] - Enhanced product assortment, including on-trend styles and categories like footwear and accessories, has fueled higher AOV, with notable sales increases in athleisure (30% year-over-year) and sneakers (35%) [4][9] - The integration of Freestyle and Fix has improved client engagement, allowing clients to start a Fix based on items discovered on the Freestyle platform [5] Group 2: Client Engagement and Personalization - The flexibility offered by larger Fixes enables clients to refresh their wardrobes according to seasonal trends and personal events, enhancing the overall customer experience [3] - The launch of themed Fixes for specific occasions has further strengthened the service's appeal, helping the company understand new customers' preferences from the outset [3] Group 3: Financial Performance - Revenue per active client increased to $542, reflecting a 3.2% rise from the previous year, attributed to strategic investments in pricing optimization and inventory management [6] - The company anticipates tougher year-over-year comparisons in fiscal 2026 but remains confident in maintaining AOV growth through continued focus on client engagement and product innovations [7] Group 4: Valuation Metrics - SFIX is currently trading at a forward 12-month price-to-sales (P/S) multiple of 0.39X, significantly lower than the industry average of 1.69X and below its median P/S level of 0.41X over the past year [8]

AI-Driven Evolution of Search - Google is focusing on an AI-forward experience, aiming for a more natural and proactive user interaction [2] - The company envisions AI integration across various Google services, creating an agent that can access and utilize user data to provide personalized assistance [2][3] - Google anticipates the evolution of its search homepage in surprising ways, with AI playing a significant role [1] Proactive and Personalized User Experience - The company aims to provide proactive assistance, such as reminding students of homework and pre-packaging relevant information [2] - Google is working towards an agent that can see across all of a user's data to provide personalized and timely information [3] Integration and Accessibility - Google emphasizes the importance of experiencing the AI-driven features firsthand, highlighting the integration across different devices [3]

Core Argument - The future of education requires a shift from content memorization to competency development, focusing on critical thinking, creative thinking, collaboration, communication, and compassion [1] - Education should embrace personalization, catering to individual learning styles and needs, aided by technology but not replacing teachers [1] - Integrated learning, breaking down subject silos, is crucial to address real-world problems that span multiple disciplines [1] - Technology should be used to empower students to become creators, not just consumers, while preserving human relationships in learning [1] - Schools need to foster innovation by creating an ecosystem that encourages inquiry, curiosity, collaboration, teamwork, digital literacy, global perspectives, resilience, and adaptability [1] Key Recommendations for Stakeholders - Teachers should focus on developing students' potential beyond knowledge acquisition [1] - Parents should advocate for education that nurtures their child's potential, not just test scores [1] - Policymakers should create assessment systems that measure what matters, not just what is easy to measure [1] - Students should understand that education is something they create for themselves [1] - Society should encourage educational innovations that cultivate wisdom, not just knowledge [1]

AI Marketing Evolution - AI is integrated into marketing to help partners grow faster and achieve goals [1] - The industry is moving towards hyper-personalization, using AI to understand consumer habits and tailor experiences [7][8][9] - AI-driven campaigns are evolving from fragmented applications to unified, end-to-end management [11][12][13] - Dynamic websites will use AI to recognize user behavior and actively close sales [17] - Smart conversation AIs will become comprehensive sales assistants, offering personalized product presentations and follow-ups [18] Ethical Considerations - The industry faces consumer distrust regarding personal data, emphasizing the need for ethical targeting and platform transparency [20][21] - Platforms should explain why a user is targeted for a specific message, avoiding biased outcomes [21][22] - Internal bias detectors and a comprehensive regulatory framework are needed to prevent discriminatory practices [23] Future Challenges and Solutions - Current AI systems lack general logic and common sense, hindering their ability to understand complex business dynamics [25][26] - Achieving general intelligence requires vast amounts of data and energy, potentially necessitating new energy sources [28][29] - AI is not an original creator and relies on original content for data, driving research into AI models mimicking the human brain [30][31] - Neural augmentation or brain-computer interfaces may be necessary to incorporate human values and address AI's limitations in understanding nuance [33][34][35]