VIPPlay Deploys Real-Time AI to Cut Sports Betting Acquisition Costs

Real-Time AI at the Edge: How Data Velocity Is Redefining Consumer Engagement With VIPPlay Inc — Ai4

Ai4

Legacy betting operators face shrinking moats and rampant churn by treating fans as transactional wallets, making real-time conversational AI essential to lower acquisition costs and drive revenue.

AI-assisted summary of Real-Time AI at the Edge: How Data Velocity Is Redefining Consumer Engagement With VIPPlay Inc — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published on YouTube by Ai4 on August 27, 2025, this presentation features Les Ottolenghi, CEO of VIPPlay Inc. According to the video description, the talk provides digital strategy and C-suite leaders with a practical blueprint for deploying real-time artificial intelligence at the edge, shifting consumer engagement from passive interactions to reactive, instant experiences across gaming and digital media.

## TL;DR Les Ottolenghi argues that legacy sports gaming and betting interfaces are fundamentally broken, alienating modern fans with static grids, complex spreads, and transactional friction that drive up churn and customer acquisition costs. In response, VIPPlay is deploying an AI-native conversational platform powered by multi-LLM routing, dynamic college football hubs, 3D spatial tailgate mapping, and personalized synthetic podcasts. The platform relies on a three-tiered real-time service layer capable of millisecond event responses and sub-second data queries supported by Decentral AI compute infrastructure. By mapping engagement across a 4x4 persona-pillar matrix, VIPPlay demonstrates how real-time behavioral personalization can drive a 10% to 15% revenue lift and lower customer acquisition costs by roughly 28%.

## Atmosphere & style The presentation carries the sharp, authoritative cadence of an executive keynote, moving at a brisk pace through market diagnostics, system architectures, and live product roadmaps. Ottolenghi speaks with the pragmatism of a seasoned technology turnaround veteran, grounding forward-looking generative AI concepts in enterprise EBITDA metrics and operational track records. The overall tone is urgent yet optimistic, evoking the feeling of an insider briefing on the edge-computing inflection point set to disrupt legacy digital entertainment.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) The competitive moat for digital platforms is shifting from transactional app navigation to AI-native conversational search and immediate answers. - [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) According to Gartner, 30% of global search traffic has migrated from conventional search engines to large language models, disrupting traditional discovery models. - [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) Legacy sports betting and gaming apps alienate casual fans by treating them as transactional wallets through cluttered number grids and confusing betting terminology. - [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) Between 60% and 70% of sports fans value social belonging over pure transactions, and 70% to 80% of younger demographics demand real-time, personalized analytics. - [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) Younger consumers overwhelmingly favor AI-first experiences, with over 70% preferring an AI interface to a standard app and 80% preferring AI over human agents. - [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) VIPPlay's conversational architecture replaces standard application navigation with a multi-LLM prompt bar routing live contextual sports data in real time. - [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) Hype Studio introduces prompt-driven micro-apps like AI-refereed smack talk and synthetic behavioral avatars to boost social fan engagement. - [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) Strategic integrations with NewRun for 3D tailgate mapping and Vocal AI for custom synthetic audio podcasts expand fan interactions beyond the screen. - [09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) VIPPlay's real-time service layer orchestrates millisecond flash-bonus delivery, sub-second query streaming, and resilient multi-model compute via Decentral AI. - [10:47](https://www.youtube.com/watch?v=YqDF7_owGGk&t=647s) The 4x4 persona-pillar matrix algorithmically aligns four user types (Avid, Fantasy, Loyalty, Casual) with four core pillars (Loyalty, Social Identity, Stats, Convenience). - [10:47](https://www.youtube.com/watch?v=YqDF7_owGGk&t=647s) Replacing batch marketing with real-time agentic workflows drives a 10-15% revenue lift, 20-30% higher retention, and an estimated 28% reduction in customer acquisition costs.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) The Airbnb Moment: How AI is Automating Legacy Moats

Opening his address by praising a preceding session delivered by Zach from customer data platform Tealium, Les Ottolenghi stresses that winning in modern enterprise requires combining artificial intelligence with real-time data streaming. Drawing directly from his personal experience as an investor and seasoned hospitality veteran, Ottolenghi recounts listening closely to a recent Airbnb earnings call to gauge how leading tech platforms are confronting rapid shifts in artificial intelligence. Roughly fifty minutes into that call, Airbnb CEO Brian Chesky delivered an insight that set the tone for the industry's broader dilemma:

> "Airbnb can't just be an app. It has to become an AI native platform. Because the new battlefield is all about search and answers. It's no longer about transactions. Guests don't want to endlessly scroll through apps. They want the right experience and they want it right now."

The reaction from institutional investors was equally telling. Wall Street analysts noted that Airbnb's traditional defensive moat—historically built around operational complexity, supply logistics, and discovery search filters—is shrinking at an unprecedented pace as generative AI automates those barriers away. Ottolenghi validates this concern by pointing to research from Gartner indicating that 30% of global search activity has already migrated away from Google toward large language models. For Ottolenghi, this represents an inflection point: the gaming, sports entertainment, and digital engagement sectors are facing the exact same structural rupture, requiring companies to disrupt their legacy offerings before competitors do it for them.

- **Global Search Disruption:** 30% of search volume has transitioned from conventional engines like Google to LLM-powered interfaces (per Gartner). - **Core Paradigm Shift:** The consumer interface is migrating from transactional scrolling across rigid screens to instantaneous, conversational search and direct answers.

---

[03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) The Broken Sports Betting Paradigm and the High Cost of Treating Fans Like Wallets

Throughout his career, Ottolenghi has leveraged structural tech transitions to build enterprise scale—from launching pioneering online travel reservation systems in 1994 and contributing to the algorithmic architecture behind the Kayak search engine, to scaling mobile gaming in Asia to deliver more than $5 billion in EBITDA. His past initiatives also include constructing a $4.4 billion social gaming infrastructure that helped steer Caesars Entertainment out of bankruptcy, as well as overhauling the digital education stack at Stride (formerly K12), propelling its equity valuation from $21 to $157 per share over a 37-month run. Stepping in as CEO of VIPPlay, he assembled an innovation team to prevent the sports entertainment business from succumbing to technical obsolescence.

The traditional online sports betting and gaming ecosystem remains fundamentally impaired by fragile, legacy architectures that reduce the consumer experience to static grids of numbers and confusing terminology. Average fans—such as a Georgia Bulldogs supporter simply attempting to back their college team—find themselves overwhelmed by cluttered interfaces and arcane wagering structures like point spreads, leading to frustrating losses. This disjointed design alienates casual users, degrading them into mere transactional "wallets" rather than valued members of a community. Consequently, operators endure punishing customer acquisition costs (CAC) and rampant churn that erode profitability.

```mermaid flowchart LR accTitle: Broken Legacy Funnel vs AI-Driven Fan Retention subgraph Legacy["Legacy Betting Ecosystem"] direction TB L1["Cluttered Grid UI\n& Complex Spreads"] --> L2["Fan Feels Treated\nLike a Wallet"] L2 --> L3["High Churn &\nSkyrocketing CAC"] end subgraph Modern["VIPPlay Real-Time Model"] direction TB M1["Conversational AI &\nSocial Belonging"] --> M2["Instant Multi-LLM\nPersonalized Context"] M2 --> M3["28% Lower CAC &\n10-15% Revenue Lift"] end ```

Yet the underlying demand from digital natives tells an entirely different story. Fans are hungry for communal connection and automated intelligence, presenting massive opportunities for operators capable of moving at the speed of real-time data:

- **60% to 70%** of sports fans report that social belonging, community connection, and shared team spirit are their primary motivators. - **70% to 80%** of Gen Z and millennial consumers expect immediate, customized delivery of sports news, player analytics, and interactive fantasy toolkits. - **71%** of younger demographics prefer interacting directly with artificial intelligence over browsing a standard mobile app interface. - **80%** would actively choose an AI interface over speaking with a human representative. - **Vast Value Creation:** Ottolenghi contrasts the stagnant margins of legacy gaming operators with modern AI-first leaders, highlighting OpenAI's run-rate approaching $12 billion in ARR and Anthropic's unit economics yielding roughly $211 per customer.

---

[06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) Product Rollouts: Hyper-Personalized College Football, Tailgate 3D Mapping, and Vocal AI Podcasts

To capture this consumer shift, VIPPlay is deploying a product roadmap focused on immersive, real-time fan interactions. Scheduled for release in September, VIPPlay's flagship college football platform—tailored initially for Tennessee Volunteers football fans—was developed alongside engineering partner Successive Technologies, led by co-founders Sid and Mark. The application bypasses traditional app layouts in favor of an integrated prompt bar supported by a custom GPT fine-tuned on thousands of public sports data feeds. Rather than relying on a single static model, the platform routes user queries simultaneously across multiple LLMs to assemble real-time intelligence on athlete health, tactical playbooks, weather conditions, schedules, and live social sentiment.

The fan experience includes interactive, prompt-driven micro-applications housed within VIPPlay's proprietary "vibe coding" suite, known as Hype Studio. Fans can engage in automated digital face-painting and launch AI-refereed smack-talk battles against rival fanbases. To humanize these exchanges, the interface integrates synthetic AI companions modeled after four distinct behavioral avatars, allowing the platform to adjust its tone and aesthetic to the demographic profile of the user.

Beyond the core conversational app, Ottolenghi outlines two complementary platform rollouts aimed at extending the fan footprint:

- **Real-Time Tailgate 3D Mapping:** Developed in collaboration with NewRun—creators of a dedicated large-scale AI event model—this spatial utility provides hyper-localized 3D spatial mapping for game-day gatherings. With a single conversational prompt, attendees can pinpoint where their friends are congregating, locate parking spaces, identify the liveliest tailgates, or check wait times at specific food trucks mapped over a personalized Google Maps layer. - **Dynamic Synthetic Podcasting via Vocal AI:** Announced via a strategic partnership, Vocal AI operates an audio platform powered by a proprietary small language model (SLM) calibrated exclusively for sports analytics and wagering insights. The engine compiles custom, near-instantaneous audio digests tailored to individual preferences, delivering roster updates, tactical breakdowns, and injury alerts in the user's preferred vocal timbre, cadence, and language.

---

[09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) The Real-Time Service Layer and Architectural Engine

Powering these user-facing experiences is VIPPlay's real-time service layer, an orchestration pipeline that connects live sports streaming feeds, specialized machine learning inference, and consumer interfaces. Central to the framework is a dynamic relevance scoring engine designed to assess, order, and serve wagering odds, context-aware offers, and dynamic content through three distinct computational pathways.

```mermaid flowchart TD accTitle: VIPPlay Real-Time Service Layer Architecture subgraph Ingestion["Edge Interaction & Event Streams"] A["Fan Prompts & User Signals"] B["Live Odds & Event Feeds"] end

subgraph ServiceLayer["VIPPlay Real-Time Service Layer"] C["Service 1: Experience Model\n(Flash bonuses & injuries in ms)"] D["Service 2: Query Engine\n(User/odds data in <1s)"] E["Service 3: ML Pipeline\n(Decentral AI Orchestration)"] end

subgraph Compute["AI Execution Matrix"] F["Decentral AI Engine\n(Nvidia Inception & UNLV Blackfire)"] G["Commercial & Open-Source LLMs\nConcurrent Inference"] end

A --> ServiceLayer B --> ServiceLayer E --> F F --> G ServiceLayer --> H["Adaptive Fan Experience\n(Hype Studio, Odds, Podcasts)"] ```

The underlying infrastructure relies on three core operational pipelines: 1. **Millisecond Experience Model:** A latency-critical event engine engineered to ingest immediate sports events—such as an unexpected in-game player injury or sudden shift in momentum—and instantly push out localized fan interventions, like dynamic flash betting bonuses, within milliseconds. 2. **Sub-Second Internal Data Query Service:** A query engine that concurrently polls high-throughput streams of active user states, sporting event updates, and fluctuating betting line movements, returning consolidated data in well under one second to preserve an effortless interactive experience. 3. **Decentral AI Model Orchestration:** A hardened machine learning pipeline constructed in partnership with Decentral AI, led by founder Omar El-Hilali. Born out of the Blackfire Innovation Center and supported through the Nvidia Inception program, Decentral AI grants VIPPlay specialized access to advanced compute infrastructure, concurrently routing fan requests across proprietary, commercial, and open-source foundation models to maintain uninterrupted inference resilience.

---

[10:47](https://www.youtube.com/watch?v=YqDF7_owGGk&t=647s) The 4x4 Persona-Pillar Matrix: Engineering Outcomes and Agentic Workflows

To operationalize real-time personalization at scale, VIPPlay applies a structured 4x4 behavioral matrix. Rather than relying on blunt demographic assumptions, the system categorizes users into four granular personas: - **Avid Millennials & Gen Z Fans:** Highly connected, digital-first supporters who demand social immersion, expressive identity, and immediate context. - **Fantasy Players:** Statistically driven users intensely focused on real-time metrics, individual player trajectories, predictive trends, and roster utility. - **Loyalty Fans:** Dedicated club enthusiasts primarily driven by status, gamified rewards, and exclusive brand equity. - **Casual Spectators:** Occasional viewers who prioritize frictionless accessibility, simplicity, and low-cognitive-load participation.

These behavioral profiles are evaluated against four operational engagement pillars: **VIP Loyalty**, **Social Identity**, **Real-Time Stats**, and **Convenience**.

| Engagement Pillar | Primary Focus & System Trigger | Representative User Persona Match | |:--- |:--- |:--- | | **VIP Loyalty** | Tailored status rewards, flash bonus unlocks, high-tier perks | Loyalty Fans | | **Social Identity** | Smack-talk battles, synthetic AI avatars, Hype Studio face-painting | Avid Millennials & Gen Z | | **Real-Time Stats** | Sub-second odds movements, roster injury feeds, predictive telemetry | Fantasy Players | | **Convenience** | Single-prompt tailgate mapping, voice-first SLM audio recaps | Casual Spectators |

By orchestrating these mappings algorithmically, VIPPlay transitions away from passive batch marketing and arbitrary push notifications, using agentic workflows that adapt on the fly to individual consumer behaviors. Modeling and stress-testing simulations demonstrate that aligning the proper operational pillar with a targeted fan persona produces measurable gains: - **10% to 15%** gross revenue expansion. - **20% to 30%** sustained improvement in customer retention. - **Approximately 28%** decrease in customer acquisition costs.

Closing his remarks, Ottolenghi re-emphasizes that sustainable growth in digital sports entertainment hinges on converting real-time behavioral telemetry into immediate personalization. By moving past static interfaces, VIPPlay aims to seize its own "Airbnb moment" and establish a new benchmark for automated fan engagement.

## Notable quotes > [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) Airbnb can't just be an app. It has to become an AI native platform. Because the new battlefield is all about search and answers. It's no longer about transactions. Guests don't want to endlessly scroll through apps. They want the right experience and they want it right now. — Brian Chesky (quoted by Les Ottolenghi)

## Mentioned Les Ottolenghi, VIPPlay Inc, Ai4, Tealium, Zach, Airbnb, Brian Chesky, Gartner, Google, Kayak, Caesars Entertainment, Stride, Georgia Bulldogs, OpenAI, Anthropic, Tennessee Volunteers, Successive Technologies, Sid, Mark, Hype Studio, NewRun, Google Maps, Vocal AI, Decentral AI, Omar El-Hilali, Blackfire Innovation Center, Nvidia Inception

## Video description Les Ottolenghi, CEO, VIPPlay Inc.

Abstracts

This presentation will explore how real-time artificial intelligence is shifting consumer engagement from passive to predictive, and now to reactive. In a world where batch processing is obsolete, I’ll explain how businesses can build systems that adapt instantly to user behavior. The talk will illustrate the “information refinery” model and how to architect for velocity across gaming, media, retail, and beyond. Attendees will learn how to design AI systems that move intelligence to the edge, closer to where decisions are made and value is created. If you're in the C-suite or leading digital strategy, this talk will give you a practical blueprint for competing at the speed of data.

[Source](https://www.youtube.com/watch?v=YqDF7_owGGk)

中文简报

VIPPlay部署实时AI以降低体育博彩获客成本 传统博彩运营商因将球迷单纯视为交易钱包而面临护城河缩小与严重用户流失，这使得实时对话式AI成为降低获客成本并推动营收的关键。 Les Ottolenghi指出，传统体育游戏与博彩界面已从根本上失效，其静态网格、复杂的让分盘口和交易摩擦疏远了现代球迷，从而推高了流失率与获客成本。为此，VIPPlay正在部署一个AI原生对话式平台，该平台由多LLM路由、动态大学橄榄球枢纽、3D空间车尾聚会地图以及个性化合成播客提供支持。平台依托Decentral AI算力基础设施的三层实时服务架构，能够实现毫秒级事件响应与亚秒级数据查询。通过在4x4画像-支柱矩阵中映射用户互动，VIPPlay展示了实时行为个性化如何推动10%至15%的营收增长，并降低约28%的获客成本。 数字化平台的竞争护城河正从交易型应用操作转向 AI 原生对话搜索与即时解答；Gartner 数据显示，全球已有 30% 的搜索流量从传统搜索引擎转向大语言模型。 (0:04) 传统体育博彩应用充斥着复杂数据与术语，将球迷异化为交易“钱包”，但数据显示 60% 至 70% 的球迷更看重社交归属感，且超 70% 的年轻受众更偏好直接与 AI 界面交互。 (3:00) VIPPlay 用多 LLM 实时路由的对话式 Prompt 栏替代传统导航，并通过 Hype Studio 引入 AI 裁判互呛与合成化身，同时整合 NewRun 3D 派对地图和 Vocal AI 定制播客拓展互动场景。 (6:29) 借助毫秒级实时服务层与 4x4 行为画像-支柱矩阵，VIPPlay 将营销转向即时自适应工作流，促使获客成本（CAC）降低约 28%，并实现 10% 至 15% 的营收增长。 (10:47)

Reading summary

VIPPlay CEO Les Ottolenghi contends that sports betting and digital entertainment face an existential crisis driven by rigid, transactional interfaces that alienate consumers. To counteract mounting acquisition costs and high churn, VIPPlay is deploying an AI-native platform designed to deliver conversational interaction, edge intelligence, and real-time hyper-personalization. The Obsolescence of Legacy Transactional Moats (0:04) Drawing parallels to Airbnb CEO Brian Chesky's declaration that platforms must transition from transactional apps to AI-native search and answers, Ottolenghi argues that traditional defensive moats in digital consumer engagement are rapidly eroding. He cites Gartner research showing that 30% of global search queries have migrated from conventional search engines to large language models, indicating a fundamental consumer shift toward instant, conversational answers. In online sports betting, legacy systems present static odds grids and complex spreads that reduce users to transactional wallets, resulting in escalating customer acquisition costs and churn. Survey metrics highlighted by Ottolenghi indicate that 60% to 70% of sports fans prioritize social belonging, while 70% to 80% of Gen Z and millennial consumers expect immediate, customized analytics. Furthermore, 71% of younger users prefer interacting with artificial intelligence over browsing conventional mobile apps, and 80% would choose an AI interface over a human representative. Conversational Interfaces and Generative Fan Utilities (6:29) To adapt to younger demographics, VIPPlay developed an AI-first college football application alongside engineering partner Successive Technologies, focusing initially on Tennessee Volunteers fans. The application bypasses conventional navigation menus in favor of a conversational prompt bar that routes questions across multiple LLMs to aggregate intelligence on weather, athlete health, team tactics, and social sentiment. User engagement is further driven by the Hype Studio suite, which provides AI-refereed smack-talk battles and synthetic AI companion avatars tailored to distinct user personas. VIPPlay's product rollout also extends into localized and auditory formats. A partnership with NewRun introduces real-time 3D spatial mapping on game days, enabling fans to identify friend locations, open parking, and food truck lines through conversational prompts overlaid on Google Maps. Additionally, an integration with Vocal AI utilizes a proprietary sports-calibrated small language model to generate dynamic, personalized audio podcasts that deliver betting analytics and roster updates in user-selected vocal timbres and languages. Real-Time Service Architecture and Model Orchestration (9:04) VIPPlay's user interfaces are supported by a multi-tier service layer that processes live sports streams and behavioral telemetry. A millisecond-latency experience model ingests in-game occurrences, such as injuries or sudden momentum changes, to immediately trigger localized interventions like dynamic flash bonuses. Concurrently, a sub-second query engine monitors high-throughput streams of active user states and shifting betting lines, returning consolidated updates in under one second. To manage continuous inference without performance degradation, VIPPlay partnered with Decentral AI, an initiative developed at the UNLV Blackfire Innovation Center and supported by the Nvidia Inception program. Led by Omar El-Hilali, Decentral AI provides a compute pipeline that concurrently routes fan prompts across open-source, commercial, and proprietary foundation models to preserve operational resilience. Behavioral Personas and Projected Unit Economics (10:47) VIPPlay operationalizes personalization through a 4x4 matrix that cross-references four fan personas—Avid Millennials & Gen Z, Fantasy Players, Loyalty Fans, and Casual Spectators—against four operational engagement pillars: VIP Loyalty, Social Identity, Real-Time Stats, and Convenience. This matrix replaces broad push notifications with agentic workflows that adapt dynamically to individual telemetry. According to VIPPlay's modeling and stress-testing simulations, aligning targeted personas with corresponding operational pillars generates significant economic advantages. The company projects a 10% to 15% expansion in gross revenue, a 20% to 30% sustained improvement in customer retention, and an estimated 28% reduction in customer acquisition costs. Ottolenghi concludes that sports entertainment operators must replace legacy transactional software with real-time, conversational AI to remain viable. However, the commercial sustainability and actual efficiency gains of VIPPlay's architecture remain anchored to initial rollouts and simulated models rather than long-term operational track records.

精炼阅读摘要

VIPPlay首席执行官Les Ottolenghi指出，传统体育博彩与数字娱乐因僵化界面及交易导向导致用户严重流失，行业正面临类似Airbnb的技术拐点；企业必须结合边缘实时数据流与多大语言模型（LLM）编排，向以即时交互和社群认同为核心的AI原生平台转型。 传统博彩模式困境与AI交互范式转变 (0:04) Ottolenghi引用Airbnb首席执行官Brian Chesky在财报电话会上的研判指出，消费级交互的主战场正在从被动翻页浏览转变为即时搜索与问答。根据Gartner研究数据，全球已有30%的搜索行为从Google等传统引擎迁移至大语言模型。这一范式转变对体育娱乐业构成了相同的技术冲击。在传统在线体育博彩中，过时的底层架构将用户体验降级为密密麻麻的数字表格与晦涩术语（如让分盘），导致普通球迷产生挫败感，并将消费者物化为单纯的“钱包”，由此产生极高的客户获取成本（CAC）与严重的客户流失率。 年轻一代数字原住民的需求呈现出截然不同的趋势。统计数据显示，60%至70%的体育迷将社交归属感、社群连接与队魂认同视为核心诉求；70%至80%的Z世代与千禧一代期望获得即时定制的体育资讯、球员动态与梦幻体育分析工具。此外，71%的年轻受众更倾向于直接与人工智能交互而非浏览常规移动应用，80%的用户在服务场景中主动选择AI界面而非人工客服。Ottolenghi认为，传统运营商利润率停滞不前，唯有向AI原生形态转型才能突破既有壁垒。 面向球迷场景的端到端生成式产品矩阵 (6:29) 为顺应消费需求变化，VIPPlay联合工程伙伴Successive Technologies开发了一套以即时交互为核心的产品体系，其旗舰级大学橄榄球应用计划于9月上线并首发服务田纳西大学志愿者队球迷。该平台取消了传统应用的多级菜单，采用单一提示词交互栏，底层搭载根据数千个公开体育数据流微调的定制GPT，能够并发调用多个大模型以实时聚合伤病状况、战术手册、天气以及社交舆情。 在核心应用之外，VIPPlay还布局了配套的交互工具与扩展功能。其自研的“Hype Studio”提供数字面部彩绘、AI裁判互动对骂，并配备基于四种行为特征的合成AI化身，以根据用户画像自适应调整交互语气。针对现场观赛群体，VIPPlay与NewRun合作推出3D尾门宴实时地图，用户输入单句提示词即可在Google Maps图层上定位好友位置、寻找车位或查询餐车排队时长；同时，平台与Vocal AI建立合作，利用针对体育分析校准的专用小语言模型（SLM），按用户偏好的音色、语速和语言即时生成定制化的赛事音频播客。 毫秒级服务层与去中心化算力编排 (9:04) VIPPlay通过构建三层实时服务层架构，将流媒体数据、机器学习推理与客户端交互无缝打通。第一层是毫秒级体验模型，专门针对突发伤病或比赛势头反转等关键节点，在数毫秒内触发精准干预措施（如推送动态闪烁投注红利）；第二层为亚秒级内部数据查询引擎，能够在远低于1秒的时间内并发轮询用户实时状态、赛事进展与赔率变动，确保前端交互平滑流畅。 第三层则是与Decentral AI联合搭建的机器学习模型编排管道。该方案依托内华达大学拉斯维加斯分校Blackfire创新中心和Nvidia Inception加速计划的算力资源，支持将用户请求并发分发至专有模型、商业模型和开源基础大模型。这种多模型动态协同架构不仅提升了高负载下的计算吞吐能力，也避免了单点故障，保障了实时推理的高可用性。 4x4行为矩阵与智能体业务成效 (10:47) 为了在规模化运营中实现精准个性化，VIPPlay设计了“4x4行为矩阵”，将用户解构为四大画像：注重社群沉浸与身份表达的千禧与Z世代狂热球迷、关注实时数据指标与球员趋势的梦幻体育玩家、重视等级奖励与专属权益的忠诚球迷，以及偏好低认知负担的休闲观众。这些画像分别与四大运营支柱进行动态算法匹配，即社交认同、实时数据、VIP忠诚度与便捷体验。 系统依靠自主智能体工作流（Agentic Workflows）依据实时行为遥测数据动态调整策略，彻底替代了传统的批量营销和粗放式推送。Ottolenghi援引压力测试与模拟数据指出，将精准运营支柱与特定用户画像动态对齐后，预计可带来10%至15%的总营收增长，使用户留存率持续提升20%至30%，并使整体客户获取成本降低约28%。 Ottolenghi总结指出，体育娱乐产业的可持续增长取决于能否将高频边缘数据转化为即时个性化体验；尽管模拟测试显示其实时架构能够显著优化获客成本与留存收益，但该方案在大规模真实赛事并发环境下的稳定运行与商业兑现能力，仍有待产品正式发布后的市场检验。

Detailed notes — zh

## Background 本内容由 Ai4 于 2025 年 8 月 27 日发布在 YouTube 平台上。视频记录了 VIPPlay 首席执行官莱斯·奥托伦吉（Les Ottolenghi）的主题演讲，主要探讨实时人工智能如何推动消费者互动模式从被动、预测走向即时反应，并分享了在边缘端构建高速数据与 AI 系统的实践蓝图。 ## TL;DR 奥托伦吉指出传统应用与复杂筛选界面的护城河正在被 AI 迅速抹平，唯有结合实时数据与 AI 原生体验才能构筑新壁垒。针对传统在线体育博彩界面晦涩、把玩家物化为钱包等痛点，VIPPlay 推出了涵盖智能 Prompt、线下 3D 尾箱派对导航以及定制 AI 播客的产品矩阵。该体系依托毫秒级响应的实时服务层与 Decentral AI 多模型协同调度，将千禧一代、范特西玩家等不同画像人群精准匹配至社交认同、即时数据等核心支柱。模拟测算显示，这种由实时信号驱动的智能体工作流可显著提升营收与留存率，并大幅降低获客成本。 ## Atmosphere & style 整场演讲充满硅谷老将特有的沉稳与行业洞察力，既有对 Airbnb 财报等前沿趋势的宏观剖析，又有直击博彩业痛点的辛辣点评。演讲节奏紧凑、逻辑严密，通过大量实战案例与架构图层层递进，传递出迫切求变的技术变革浪潮。现场氛围既具商业说服力，又展现出对年轻一代球迷文化的深刻共情与创新热情。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 人工智能与实时数据的深度融合正在重构各行业的竞争壁垒，推动业务从传统的批量处理与被动交互全面转向即时响应。 - [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 数字化竞争重心已从海量信息浏览与交易转向精准即时的“搜索与解答”，平台必须向 AI 原生架构转型以避免被边缘化。 - [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 传统企业依靠产品复杂性、筛选器和物流建立的护城河正被 AI 自动化抹平，全球已有约 30% 的搜索流量转移至大语言模型。 - [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) 传统体育博彩行业存在界面晦涩拥挤、规则缺乏解释以及过度将用户工具化等软肋，造成高昂的获客成本与脆弱的用户留存。 - [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) 年轻一代呈现出显著的代际需求迁移，超过 70% 的 Z 世代与千禧一代更倾向于直接与 AI 交互，追求社交归属感与轻量化体验。 - [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) VIPPlay 打造 AI 原生球迷生态，通过“Hype Studio”氛围编程与智能 Prompt 交互栏取代繁琐点击，提供虚拟彩绘与对抗对决等趣味玩法。 - [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) 结合 3D 空间渲染与自然语言 Prompt，AI 可高效赋能线下 Tailgate 赛前聚会场景，秒级解决找车位、寻美食与好友定位问题。 - [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) 依托专用小语言模型（SLM），平台能够根据每位用户的球队偏好与投注习惯，动态合成千人千面的个性化体育分析播客。 - [09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) 核心底座“实时服务层”整合了毫秒级突发事件模型、亚秒级数据查询以及 Decentral AI 的多模型并发编排管线，保障极致响应速度。 - [09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) 基于实时行为信号驱动的智能体工作流（Agentic Workflows），将细分用户画像与核心体验支柱精准匹配，能有效提升留存并降低约 28% 的获客成本。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 实时数据与 AI 的交汇：从 Airbnb 财报看各行业的护城河重塑 VIPPlay 首席执行官莱斯·奥托伦吉（Les Ottolenghi）以呼应前一位演讲者（Tealium 的扎克）开场，指出结合“人工智能与实时数据”的洞见不仅极具价值，更是一切行业变革的核心基石。他回顾了自己上周收听 Airbnb 财报电话会议的经历——出于深厚的酒店业背景、作为股东的关注，以及对这家创新企业如何应对 AI 浪潮的强烈好奇。会议进行到约 50 分钟时，Airbnb 首席执行官布莱恩·切斯基（Brian Chesky）的一番话令他印象极深：“Airbnb 不能仅仅是一个应用程序，它必须成为一个 AI 原生平台（AI-native platform）。”切斯基强调，数字化竞争的新战场已经彻底转向“搜索与解答”，而非单纯的“交易”；用户不再愿意在 App 里漫无目的地无限滑动，他们需要精准且即时的体验。 更让奥托伦吉感到震撼的，是机构投资者的敏锐反馈。一位资深投资人一针见血地指出：“护城河正在缩小。曾经保护 Airbnb 的护城河——曾经构建起极高进入门槛的复杂性、物流与搜索筛选器——正在被 AI 逐步自动化抹平。如果 Airbnb 不迅速行动，它的护城河将荡然无存。”奥托伦吉深表认同，并引用了 Gartner 的权威报告作为佐证：全球已有 30% 的搜索流量正在从 Google 转移至各类大语言模型（LLM）。在体育娱乐与在线博彩领域，同样的“Airbnb 时刻”和“Google 时刻”已然降临，唯有将 AI 与实时数据紧密结合的颠覆者，才能在重塑自身业务的同时赢得未来。 - **Gartner 行业数据**：全球已有 30% 的搜索流量从 Google 转移到大语言模型（LLM）。 - **切斯基核心论点**：App 时代落幕，AI 原生平台崛起，用户需求从海量信息浏览转为即时问答与精准体验。 - **资本市场预警**：传统平台依赖的产品复杂度与筛选逻辑构筑的壁垒正被自动化消解，缺乏 AI 创新的业务面临失去竞争优势的风险。 --- ### [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) 在线博彩与体育娱乐业的痛点与代际迁移 作为一名连续利用前沿技术重塑产业的技术老将，奥托伦吉历数了自己的职业生涯轨迹：从 1994 年率先推出首个在线旅游预订服务、发明 Kayak 搜索引擎，到在亚洲落地移动游戏业务创造逾 50 亿美元的 EBITDA；从为凯撒娱乐（Caesars Entertainment）打造价值 44 亿美元的社交游戏平台以助其摆脱破产困境，再到近期应用数据与 AI 将教育平台 Stride（K12）的股价在 37 个月内从 21 美元拉升至 157 美元。上任 VIPPlay 首席执行官后，他组建了一支实力过硬的创新团队，决心在行业面临掉队危机之际，主动把握这场属于体育博彩业的变革时刻。 奥托伦吉直言不讳地揭露了当前传统在线体育博彩行业的系统性软肋：技术架构陈旧脆弱、系统异常复杂，最严重的是“将用户仅仅当成一个钱包”。用户面对的是塞满各种晦涩方块、赔率术语与文字的杂乱屏幕。他举例道，一个普通的佐治亚斗牛犬队（Georgia Bulldogs）球迷可能只想简单支持自己的主队，却因为完全不懂复杂的“让分盘（spread）”规则而莫名其妙输钱。这种体验导致客户获取成本（CAC）高居不下、客户流失率吞噬利润，且用户毫无忠诚度可言。然而，新一代年轻群体正释放出截然不同的需求信号：他们渴求认同感、轻量化与智能化互动，行业若固步自封，必将被边缘化。 - **高管行业履历**： - 1994 年首创在线旅游预订系统，发明 Kayak 搜索引擎； - 落地亚洲移动游戏平台，带来超过 50 亿美元 EBITDA； - 为 Caesars 打造 44 亿美元社交游戏业务，助力企业走出破产清算； - 操刀 Stride Learn 平台 AI 架构，37 个月内驱动股价由 21 美元攀升至 157 美元。 - **传统博彩行业困局**：界面拥挤晦涩（让分盘等机制缺乏解释）、把玩家物化为提款机、获客成本高企、客户留存脆弱且缺乏网络效应。 - **新世代用户画像数据**： - 60% 至 70% 的体育迷看重球队归属感与社交连接； - 70% 至 80% 的 Z 世代及千禧一代期望便捷、高度个性化的资讯、球员数据与范特西（Fantasy）工具； - 71% 的年轻受众更倾向于直接与 AI 互动而非操作死板的 App； - 更有 80% 的年轻受众在特定场景下更偏好与 AI 交互而非人类客服。 - **前沿 AI 商业标杆**：OpenAI 年化经常性收入（ARR）达到 120 亿美元，Anthropic 亦展现出单客 211 美元的健康盈利能力。 --- ### [06:29](https://www.youtube.com/watch?v=YqDF7_owGGk&t=389s) VIPPlay 产品矩阵：以 AI 为原生的球迷互动生态 针对上述痛点，VIPPlay 计划于 9 月推出首款面向大学橄榄球（College Football）的 AI 原生应用，首发版本专为田纳西大学志愿者队（Tennessee Volunteers）球迷定制。该系统由 Successive Technologies 团队（其创始人 Mark 与 Sid 亦在现场）联合打造，底层训练汇聚了数千个公共数据源，并依托多模型协同实时生成关于赛程、即时比分、伤病动向与社交媒体热梗的洞察。产品抛弃了复杂的点击层级，配备智能 Prompt 栏与基于社区热搜的引导提示。依托其名为“Hype Studio”的“氛围编程（vibe coding）”平台，应用推出了 AI 虚拟面部彩绘、球迷隔空互喷（smack talk battles）等趣味互动，并根据粉丝画像量身提供 4 类各具风格的虚拟数字化身（Avatar）。 ```mermaid flowchart TD accTitle: VIPPlay 实时服务架构与产品流向图 accDescription: 展示多源数据与模型如何经由实时服务层处理并分发至三大前端产品 subgraph DataAndModels["多源输入与编排层"] A["多 LLM / SLM 引擎<br/>(Decentral AI 协同调度)"] B["多源实时数据流<br/>(赔率 / 赛事实时流 / 行为信号)"] end subgraph ServiceLayer["VIPPlay 实时服务层 (核心大脑)"] C["即时体验模型<br/>(毫秒级突发事件/限时奖励)"] D["内部实时数据查询<br/>(亚秒级响应)"] E["智能相关性推荐引擎<br/>(内容 / 赔率动态排序)"] end subgraph EndProducts["前端产品矩阵"] F["大学橄榄球 AI 平台<br/>(Hype Studio / 智能 Prompt)"] G["3D 线下集聚社区<br/>(NewRun 尾箱派对导航)"] H["Vocal AI 专属播客<br/>(个性化小语言模型 SLM)"] end A --> ServiceLayer B --> ServiceLayer C --> E D --> E E --> F E --> G E --> H ``` 除首款旗舰应用外，奥托伦吉还揭晓了另外两大战略级产品。第二款是携手拥有最大 AI 活动事件模型的 NewRun 打造的实时社区工具，专注于大学橄榄球赛前聚会（Tailgates）。该功能在 Google 地图上叠加实时 3D 渲染，用户只需输入自然语言 Prompt，即可在几秒内获悉好友当前方位、空余车位分布、口碑最佳的尾箱烧烤点或流动餐车。第三款则是与 Vocal AI 联合发布的体育博彩 AI 定制播客平台。该平台搭载专用小语言模型（SLM），能根据每位用户的球队偏好、投注风格与语言习惯，以用户偏爱的语调、人声和节奏，动态自动生成包含最新情报与赛事分析的个性化专属音频。 - **首发产品亮点（9 月上线）**： - 聚焦田纳西大学橄榄球队，多模型跨模态并行处理千余种公共数据源； - 内置基于社区搜索趋势的 Prompt 交互栏； - Hype Studio 驱动的轻量互动：AI 虚拟彩绘、粉丝对抗玩法与 4 种风格的虚拟形象。 - **线下场景应用（NewRun 合作）**：结合 3D 空间地图与 Google 地图底层，自然语言解决 Tailgate 找车位、找美食、找好友等复杂现场痛点。 - **动态音频平台（Vocal AI 合作）**：基于专有 SLM 自动合成高度定制的体育音频播客，支持个性化音色、语调与多语言实时转换。 --- ### [09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) 实时服务层架构与多维画像增长模型 在揭秘这套系统如何运作时，奥托伦吉将核心归结为一个强大的“实时服务层（Real-time Service Layer）”。这正是连接现场实时数据、底层 AI 模型与终端用户体验的核心枢纽。该层级的运转依赖三大核心支撑：首先是高安全性的“实时体验模型”，能在毫秒级响应突发事件（如球员临时伤退或限时红利派发）；其次是“内部实时数据查询服务”，可在 1 秒以内穿透查询比赛事件流、用户行为轨迹与动态赔率变动；最后是通过深度合作方 Decentral AI 构建的安全机器学习管线。Decentral AI 创始人奥马尔·希拉利（Omar El-Hilali）现场出席，该企业孵化自内华达大学拉斯维加斯分校的 Blackfire 创新中心，同时是英伟达初创加速计划（NVIDIA Inception）成员，能为 VIPPlay 提供商用及开源 LLM 的并发动态路由调度能力与算力基础设施支撑。 为了证明这套架构在商业端的可落地性，奥托伦吉展示了一张精密的互动热力矩阵图。VIPPlay 将用户解构成四大特征人群：热衷移动互联的千禧一代/Z世代狂热球迷、范特西（Fantasy）深度玩家、忠诚铁杆拥趸、以及泛休闲观赛者；并将其精确映射至四大体验支柱：VIP 专属忠诚度、社交身份认同、实时数据统计、以及极致便利性。系统利用实时捕获的行为信号驱动智能体工作流（Agentic Workflows），动态调整呈现的内容与赔率。奥托伦吉总结道，基于严谨的模拟测算，这种“在毫秒级将正确的体验支柱精准匹配给对应人群”的机制并非靠盲目猜测，而是通过工程化手段锁定确定性增长，真正实现规模化扩张。 | 用户画像（Personas） | 核心关注体验支柱（Key Pillars） | AI 赋能的典型互动场景 | |:--- |:--- |:--- | | **热衷移动互联的千禧/Z世代** | 社交身份认同（Social Identity） | Hype Studio 虚拟彩绘、AI 互喷对决、赛前 Tailgate 3D 地图寻友 | | **范特西深度玩家（Fantasy Players）** | 实时数据统计（Real-Time Stats） | 亚秒级球员动态追踪、伤病即时情报推送、多模型协同策略问答 | | **忠诚铁杆拥趸（Loyalty Fans）** | VIP 专属忠诚度（VIP Loyalty） | 动态专属红利发放、定制化虚拟形象匹配、高阶社群权益分配 | | **泛休闲观赛者（Casual Spectators）** | 极致便利性（Convenience） | 简化赔率认知、自然语言 Prompt 代替繁复筛选、Vocal AI 智能摘要播客 | - **三大实时后台支柱**： - **即时体验模型**：毫秒级捕获球场伤病、即时让分调整及限时奖励派发； - **查询服务**：低至 1 秒以内完成对庞大赛事流、用户画像与赔率数据的并发穿透； - **Decentral AI 协同系统**：NVIDIA Inception 成员，实现商用与开源大模型的并发混合编排。 - **业务模拟预测成效**： - 预计综合营收提升 **10% 至 15%**； - 用户留存率提升 **20% 至 30%**； - 获客成本（CAC）降低约 **28%**。 - **核心演进逻辑**：实时行为信号 $\rightarrow$ 驱动智能体工作流（Agentic Workflows） $\rightarrow$ 极致即时个性化 $\rightarrow$ 构筑沉淀忠诚度 $\rightarrow$ 形成全新商业护城河。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) Airbnb 不能仅仅是一个应用程序，它必须成为一个 AI 原生平台（AI-native platform）。 — Brian Chesky > [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 护城河正在缩小。曾经保护 Airbnb 的护城河——曾经构建起极高进入门槛的复杂性、物流与搜索筛选器——正在被 AI 逐步自动化抹平。如果 Airbnb 不迅速行动，它的护城河将荡然无存。 — 机构投资者 > [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 全球已有 30% 的搜索流量正在从 Google 转移至各类大语言模型（LLM）。 — Les Ottolenghi > [03:00](https://www.youtube.com/watch?v=YqDF7_owGGk&t=180s) 将用户仅仅当成一个钱包。 — Les Ottolenghi > [09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) 在毫秒级将正确的体验支柱精准匹配给对应人群。 — Les Ottolenghi > [09:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=544s) 实时行为信号 → 驱动智能体工作流（Agentic Workflows） → 极致即时个性化 → 构筑沉淀忠诚度 → 形成全新商业护城河。 — Les Ottolenghi > [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 数字化竞争的新战场已经彻底转向“搜索与解答”，而非单纯的“交易”；用户不再愿意在 App 里漫无目的地无限滑动，他们需要精准且即时的体验。 — Brian Chesky > [00:04](https://www.youtube.com/watch?v=YqDF7_owGGk&t=4s) 唯有将 AI 与实时数据紧密结合的颠覆者，才能在重塑自身业务的同时赢得未来。 — Les Ottolenghi ## Mentioned Les Ottolenghi, VIPPlay Inc, Brian Chesky, Airbnb, Google, Gartner, Kayak, Caesars Entertainment, Stride, Georgia Bulldogs, Tennessee Volunteers, Successive Technologies, Mark, Sid, Hype Studio, NewRun, Vocal AI, Decentral AI, Omar El-Hilali, Blackfire, UNLV, NVIDIA Inception, OpenAI, Anthropic ## Video description Les Ottolenghi, CEO, VIPPlay Inc. Abstracts This presentation will explore how real-time artificial intelligence is shifting consumer engagement from passive to predictive, and now to reactive. In a world where batch processing is obsolete, I’ll explain how businesses can build systems that adapt instantly to user behavior. The talk will illustrate the “information refinery” model and how to architect for velocity across gaming, media, retail, and beyond. Attendees will learn how to design AI systems that move intelligence to the edge, closer to where decisions are made and value is created. If you're in the C-suite or leading digital strategy, this talk will give you a practical blueprint for competing at the speed of data. [Source](https://www.youtube.com/watch?v=YqDF7_owGGk)

Source quotes

[0:04] Brian Chesky: Airbnb 不能仅仅是一个应用程序，它必须成为一个 AI 原生平台（AI-native platform）。 [0:04] 机构投资者: 护城河正在缩小。曾经保护 Airbnb 的护城河——曾经构建起极高进入门槛的复杂性、物流与搜索筛选器——正在被 AI 逐步自动化抹平。如果 Airbnb 不迅速行动，它的护城河将荡然无存。 [0:04] Les Ottolenghi: 全球已有 30% 的搜索流量正在从 Google 转移至各类大语言模型（LLM）。 [3:00] Les Ottolenghi: 将用户仅仅当成一个钱包。 [9:04] Les Ottolenghi: 在毫秒级将正确的体验支柱精准匹配给对应人群。 [9:04] Les Ottolenghi: 实时行为信号 → 驱动智能体工作流（Agentic Workflows） → 极致即时个性化 → 构筑沉淀忠诚度 → 形成全新商业护城河。 [0:04] Brian Chesky: 数字化竞争的新战场已经彻底转向“搜索与解答”，而非单纯的“交易”；用户不再愿意在 App 里漫无目的地无限滑动，他们需要精准且即时的体验。 [0:04] Les Ottolenghi: 唯有将 AI 与实时数据紧密结合的颠覆者，才能在重塑自身业务的同时赢得未来。

Source quotes

[0:04] Brian Chesky (quoted by Les Ottolenghi): Airbnb can't just be an app. It has to become an AI native platform. Because the new battlefield is all about search and answers. It's no longer about transactions. Guests don't want to endlessly scroll through apps. They want the right experience and they want it right now.

Original transcript

[0:04] Thank you for attending my talk. If you [0:06] heard the the previous talk by Zach from [0:09] Tillium, um you really heard the essence [0:12] of what we're practicing. Uh by the way, [0:15] if you didn't hear it, go by his booth. [0:17] I I think they're spoton. And that's [0:20] because uh what he said were really [0:24] really good ideas, but better than that, [0:27] they were great ideas. [0:30] And what I've heard from this conference [0:32] are a number of great ideas that are [0:34] true across all industries. [0:37] The type of ideas that combine AI with [0:40] real time data. [0:43] And that's really the foundation of [0:44] success as we see it here at VIP Play. [0:49] Last week, um, I listened to the Airbnb [0:53] earnings call and that's because I have [0:55] a long history in hospitality. [0:58] I also own the stock. And most of all, I [1:02] was curious about how this innovative [1:05] company is thinking about AI and what [1:08] they're going to do about it. [1:10] And well, about 50 minutes into the [1:13] earnings call, Brian Chesy, the CEO, [1:16] said something that stuck with me [1:19] specifically. He said, and I quote, [1:23] "Airbnb can't just be an app. [1:27] It has to become an AI native platform. [1:32] And Brian added, "Because the new [1:35] battlefield is all about search and [1:39] answers. It's no longer about [1:41] transactions. Guests don't want to [1:44] endlessly scroll through apps. [1:47] They want the right experience and they [1:49] want it right now." [1:53] But what was most interesting were the [1:55] comments from the institutional [1:56] investors. And one comment stuck out to [1:59] me. And here's the quote. The moat is [2:03] shrinking. The moat that once protected [2:06] Airbnb is shrinking. [2:09] Complexity, logistics, and search [2:12] filters used to be the barriers to [2:14] entry. Now AI is automating the moat [2:17] away. And if Airbnb doesn't act fast, [2:21] there will be no moat. [2:24] I agree with all those statements [2:26] because you can see the reality in [2:28] Gartner's report that 30% of global [2:31] search has shifted away from Google and [2:35] over to large language models. And it's [2:39] that Airbnb moment, that Google moment [2:43] where sports entertainment and online [2:44] gaming are today. [2:47] which means that the winners will [2:49] combine AI and real-time data to disrupt [2:52] their markets and disrupt their own [2:54] business. [2:57] So, that's where we're going and that's [3:00] what we're doing at VIP Play. We're [3:03] changing the business through [3:05] technology, which is something I've [3:06] always done. Whether it was introducing [3:09] the first online travel reservation in [3:11] 1994, or inventing the kayak search [3:14] engine or implementing mobile gaming in [3:17] Asia that helped yield over $5 billion [3:19] in Ibida, or for Caesar's Entertainment, [3:23] building a $4.4 billion social gaming [3:26] platform that helped the company come [3:28] out of bankruptcy. [3:30] and most recently applying data NI and [3:33] AI to build Stride Learn K12 Learnings [3:37] platform which pushed the stock from $21 [3:41] to $157 and 37 months. [3:46] So, as VIP's new CEO, I recruited a [3:49] proven innovation team to pursue our [3:51] industry's Airbnb moment while avoiding [3:54] the risk of falling behind. [3:57] But what does the risk look like? [4:00] Well, the risk of the industry looks [4:02] like old technology that's fragile and [4:07] complex [4:09] and unfortunately treats the customer [4:12] like they were a wallet. [4:15] There's no good experience, [4:17] just squares and a bunch of information, [4:21] frankly, that's hard to understand. [4:24] And if I'm the average sports fan, I'm [4:26] staring at one of these mobile screens [4:28] and it's crowded with all these squares [4:30] and words. [4:32] I'm just wanting to bet on the Georgia [4:35] Bulldogs if I'm a Georgia Bulldog fan. [4:38] But I likely lost money because when I [4:41] bet I didn't cover something called the [4:44] spread. [4:45] These problems make the economics of the [4:48] sports betting industry too high with [4:50] the cost of acquisition for a customer [4:53] or the churn eating into the [4:56] profitability [4:57] and as a result fans aren't loyal. [5:01] They're made to feel like that wallet [5:03] and not like a valued customer. [5:06] So the result is it's a weak ecosystem. [5:09] no network effects and a limit on the [5:11] size of the market which means that the [5:14] online gaming business is at risk unless [5:19] unless they seize on the AI and [5:22] real-time data moment. They capture the [5:25] 60 to 70% of the sports fans that want [5:27] to feel team spirit and social [5:30] connection. [5:32] the 70 to 80% of Gen Z and millennials [5:35] that want easy personalized access to [5:39] sports news, player info, and fantasy [5:41] tools. [5:43] Or the 71% that same demographic would [5:47] want to interact with an AI rather than [5:49] an app. Or even more astounding, the 80% [5:54] would rather interact with an AI and not [5:57] with a human. [5:59] And it's this moment, this AI moment [6:02] that it's possible to miss the [6:04] extraordinary returns. I think we all [6:07] know that OpenAI makes 12 billion in ARR [6:11] and Anthropic is profitable by measure [6:14] of $211 [6:17] per customer, [6:19] which makes you think, where was AI just [6:21] three years ago in the gaming industry? [6:24] And in three years, where will the [6:26] gaming industry be with AI? [6:29] So VIP is seizing the moment. Our first [6:34] product will be released in September [6:36] and was built by success of technologies [6:39] whose founders Mark and Sid are here [6:41] with us. [6:42] And the September release is an AI [6:45] college football platform. And in this [6:48] example, it's made for the Tennessee [6:50] college football fans. Now, the system [6:53] has a GPT trained on thousands of public [6:57] data sources with real-time insights [7:00] generated by multiple LLMs about [7:03] players, schedules, news, scores, and [7:08] social trends. It has a prompt bar, [7:11] trending prompts based on other fans [7:13] searches, and it answers questions by [7:16] pulling from multiple LLMs at once. [7:20] And the fan engagement includes these [7:23] prompt-driven apps like an AI face [7:25] painting and fan smack talk battles [7:29] which are all all part of our vibe [7:31] coding platform called hype studio. The [7:35] system also includes for per the [7:38] personal avatars selected from four [7:40] different personas matching the style of [7:43] the fan to their demographic. [7:47] Our [7:51] second product is an AI community app [7:56] powered by New Run who has built the [7:58] largest AI event model that brings [8:01] real-time 3D mapping to football [8:03] tailgates and pregame gettogethers. [8:07] Want to know where your friends are? [8:10] Where to park? Where are the best [8:12] tailgates or food trucks? One prompt [8:15] gives you all the answers in a [8:16] personalized Google map. [8:20] In addition, on Tuesday, [8:23] we announced our partnership with Vocal [8:26] AI. [8:28] That's an AI podcast platform designed [8:30] specifically for sports betting and [8:33] powered by Vocal's proprietary small [8:35] language model. [8:38] Vocal knows your team, your interests, [8:41] and delivers content in a podcast with [8:44] the latest information and news based on [8:47] your preferences and in your favorite [8:49] tone, voice, speech, or language. [8:53] We're very excited about the vocal [8:55] partnership because we think that [8:56] platform is actually the future of all [8:59] sports p sports podcasts and [9:01] experiences. [9:04] So, how will this magic work? [9:06] Well, each fan interaction gets passed [9:10] into a real-time service layer. [9:14] This is the brain that connects live [9:16] data, AI models, and fan experiences. [9:21] And the key is our relevance model that [9:23] ranks offers odds and content instantly [9:27] on three real time services. The first [9:31] service is our secure real-time [9:33] experience model that surfaces breaking [9:36] events like a player injury or a flash [9:39] bonus all within milliseconds. [9:42] The second is our internal real time [9:45] data query service that pulls from event [9:48] users and odd streams in under a second. [9:52] So the response feels instant. [9:56] And the third is a secure machine [9:58] learning pipeline connected to our [10:00] partner Decentral AI who has built an [10:03] advanced AI orchestration platform [10:07] connecting VIP concurrently to [10:09] commercial and open-source LLMs in real [10:13] time. Mar Hilali the founder of [10:16] Decentral AI is with us here today. [10:19] And yesterday we announced our [10:21] industry-leading partnership born out of [10:23] Blackfire innovation center. Decentral [10:26] AI is also part of Nvidia's inception [10:29] program and has advanced access to key [10:32] infrastructure that will provide VIP [10:35] with a competitive advantage. [10:38] And as you see from this video, the [10:41] real-time decision making is happening [10:43] and it's on a streaming basis. [10:47] So, how do we think all of this will [10:49] work out? If you take a look at this [10:52] heat map, there are four personas like [10:55] the Tennessee synthetic AI fan buddies [10:58] that represent one avid millennials, Gen [11:02] Z fans, two fantasy players, three [11:08] loyalty fans, and four casual [11:11] spectators. [11:13] These personas can be matched with four [11:15] engagement pillars. One, VIP loyalty. [11:20] Two, social identity. Three, real-time [11:24] stats. And four, convenience. [11:28] Every square shows with precision how [11:32] engaged that persona is on a specific [11:34] pillar and on a personalized AI [11:37] experience. [11:39] And here's the power. When we match the [11:42] right pillar to the right persona in [11:43] real time using AI, we're not guessing. [11:48] Just as Zach said in the previous [11:50] presentation, we're engineering outcomes [11:53] with our simulation showing that there's [11:55] a potential for a 10 to 15% revenue [11:59] lift, [12:00] 20 to 30% higher retention, and [12:03] approximately 28% lower acquisition [12:06] costs. [12:07] And that means we're doing it not just [12:10] as AI but AI infused with real- time [12:12] data looking at behavioral signals again [12:16] just as Zach had presented before my [12:18] presentation [12:20] and these are powering agentic workflows [12:22] the shifts in the behaviors that each [12:24] consumer is looking for for [12:26] personalization. [12:28] That means we can adapt to what each van [12:31] wants right now. And that [12:33] personalization will turn into loyalty [12:36] and loyalty into growth. I believe this [12:40] is how you can scale with AI and [12:41] real-time data. And this is how we're [12:44] making this opportunity our ARBNB [12:46] moment. [12:48] Thank you for attending my talk. If [12:50] you're interested in learning more or [12:52] would like to connect with me, please [12:53] scan my QR code here and learn more [12:56] about VIP and how we're using AI to win. [13:01] [Applause]