The New Jersey Devils have introduced “Bott Stevens,” a bespoke AI chatbot aimed at improving digital engagement with fans.
Named in honor of Devils icon Scott Stevens, this chatbot utilizes the decentralized AI framework of Theta EdgeCloud (THETA) and will be accessible on the franchise’s official website throughout the 2024-25 NHL season.
The AI companion will furnish fans with up-to-the-minute information regarding game schedules, ticket availability, statistics, and merchandise. Leveraging Theta’s Retrieval Augmented Generation technology, Bott Stevens will source data from official NHL channels, ensuring accuracy and mitigating the risks of misinformation from unreliable sources.
Fans will be able to pose a variety of questions to the chatbot, whether inquiring about upcoming match-ups or seeking details on season statistics, receiving prompt replies in return.
Bott Stevens will harness Theta EdgeCloud’s extensive network comprising over 30,000 edge nodes and distributed GPUs, delivering over 80 PetaFLOPS of computing power.
This capability guarantees that the chatbot can scale effectively during peak periods, such as playoffs or significant team announcements.
In addition to answering queries, the AI chatbot will provide historical highlights, game summaries, venue details, and information about team events. Future developments may include predictive analytics for fantasy sports and more interactive tools for enhancing fan engagement.
Q&A with Mitch Liu, CEO of Theta Labs
Q: How does Theta EdgeCloud’s decentralized model ensure the chatbot can manage peak demand during important events?
A: The decentralized model of EdgeCloud offers over 80 PetaFLOPS of on-demand distributed GPU computing power by utilizing both cloud and edge computing devices. With more than 30,000 edge devices worldwide, we can accommodate peak demand that surpasses the capacity typically provided by cloud servers, which is essential in professional and esports sectors, especially when events like playoffs and finals generate maximum fan interest.
Q: How does “Bott Stevens” access accurate, real-time information from sources like the NHL API? What measures are in place to avoid outdated or incorrect data? How do you prevent RAG from retrieving information from unreliable crowd-sourced sources?
A: The EdgeCloud back-end system identifies which data sources to use, directing this into the real-time RAG database to ensure that crowd-sourced or inaccurate information is excluded. A significant challenge in generative AI models is not just sourcing accurate real-time data from reliable outlets, but also how to effectively clean, prepare, and standardize that data for seamless integration into the model.
Q: How are you safeguarding user privacy and security for those interacting with the chatbot? Is user information stored?
The chatbot does not retain or store any user data beyond the duration of the session. Unlike conventional chatbots such as DeepSeek or ChatGPT, Theta’s chatbots focus solely on industry-specific content, providing information exclusively related to hockey for the New Jersey Devils.
Q: Are there plans for future enhancements to “Bott Stevens,” such as predictive analytics? How far do you foresee developing the technology?
A: We are currently focusing on expanding the interactive RAG chatbot into a proactive agentic AI capable of various independent actions. For instance, it could interface with customer service and CRM systems to automatically generate support tickets, thereby enhancing user satisfaction. Additionally, AI agents could create fantasy sports or esports teams to compete against users, thereby increasing fan engagement and paving the way for predictive analytics in areas like talent scouting, player valuation, and opponent analysis.
Q: What prompted Theta Labs to collaborate with the New Jersey Devils? What customization possibilities exist for other teams or brands?
A: Theta EdgeCloud has garnered significant attention from professional sports franchises, starting with teams such as the Las Vegas Knights, and from esports organizations like FlyQuest and Evil Geniuses. The user interface is fully customizable and can be integrated into each team’s website, mobile apps, Discord, and other social platforms. More crucially, the real-time data sources and APIs that feed into EdgeCloud’s RAG database are tailored to specific domains, such as NHL hockey and various esports titles.
Q&A with Marc Ciampa, Vice President of Content of the NJ Devils
Q: How will the team market “Bott Stevens” to boost fan engagement and adoption?
A: To promote “Bott Stevens,” the Devils plan to leverage a multi-platform marketing strategy aimed at encouraging fan interaction with the chatbot and raising awareness of its capabilities.
Q: What types of curated content will the chatbot provide beyond statistics and schedules?
A: In addition to stats and schedules, “Bott Stevens” will deliver curated content that enhances the fan experience, such as historical highlights and details about events and venues. It will respond to inquiries about memorable moments in the team’s history and provide information on upcoming events, themed nights, and comprehensive details about Prudential Center amenities, including accessibility services, bag policies, and locations of concessions and ATMs throughout the venue.
Q: How will you measure the success of the chatbot in boosting fan engagement or offering relevant and precise data?
A: We will evaluate success through metrics like user engagement, accuracy rates, feedback from fans, and the reduction of support load. We’ll track the number of interactions, session lengths, and repeat usage, along with the accuracy of information shared, ensuring that stats and data are correct and timely. Additionally, we will gather user feedback through surveys to assess satisfaction levels, identify areas for improvement, and measure the decline in inquiries managed by human support, signifying the chatbot’s efficiency in addressing common questions.
Q: As AI fatigue grows among consumers, how will you differentiate “Bott Stevens” from other AI agents that people encounter regularly?
A: By facilitating continuous learning, adopting a human-like conversational style, providing personalized interactions, and creating a unique chatbot ecosystem. Machine learning algorithms will adapt to fan feedback and changing interests, ensuring the chatbot stays relevant and engaging while its communication style will be designed to feel friendly and relatable, enhancing overall user experience. We aim to use AI to customize responses based on individual fan preferences and habits, making every interaction distinct. Eventually, “Bott Stevens” will be integrated into our app and connected to other key features such as concession guides, navigation, and more interactive tools we may develop in the future.