The New Jersey Devils have launched Bot Stevens, a custom AI chatbot designed to enhance digital fan engagement.
Named after Devils legend Scott Stevens, the chatbot features Theta Edgecloud (Theta) decentralized AI infrastructure and is available on the team’s official website during the 2024-25 NHL season.
AI agents provide fans with real-time information about game schedules, tickets, statistics and products. Using Theta’s search and enhanced generation technology, Bott Stevens extracts data from official NHL sources to ensure accuracy while avoiding false information from unverified sources.
You might like it too: Crypto Exchange Deribit uses fake recruiter profiles to warn against employment fraud
Fans can ask the chatbot a variety of questions, including details of upcoming matches and season statistics, and get instant answers.
Bott Stevens features over 30,000 edge nodes and a network of distributed GPUs, providing over 80 PetaFlops computing power.
This ensures scalability even in peak demand, such as playoffs and major team announcements.
In addition to answering questions, the AI chatbot provides historical highlights, game summary, venue information and team events updates. Over time, we may expand to include predictive analytics for fantasy sports and interactive fan engagement tools.
You might like it too: Privy has secured more than $40 million in funding to expand its crypto infrastructure
Q&A with Mitch Liu, CEO of Theta Labs
Q: How does Theta Edgecloud’s distributed design allow chatbots to handle peak demand during major events?
A: EdgeCloud’s distributed design combines both cloud and edge computing devices to provide over 80 petaflop-on-demand distributed GPU computing power. Over 30,000 global edge devices allow you to handle peak usage across the baseline capacity provided by cloud-based servers, and key events such as playoffs and finals are important in the industry that are most attracting fan interest in professional and esports industries.
Q: How does “Bot Stevens” extract accurate, real-time data from sources like the NHL API? Which safeguards prevent outdated or misinformation? How can I prevent RAG from being pulled from crowdsourced, potentially inaccurate sources?
A: EdgeCloud’s backend platform defines the data source from which it extracts data. This feeds into a real-time RAG database, so crowdsourced or inaccurate sources can be avoided. Like most generative AI LLM models, the biggest challenge is not only how to obtain accurate real-time data from reliable sources, but how that data can be cleaned, prepared, normalized and efficiently and accurately incorporated into the model.
Q: How do you guarantee data privacy and security for users interacting with the chatbot? Do you store user data?
Chatbots do not store or maintain user data beyond the user’s sessions, and unlike traditional chatbots such as DeepSeek and ChatGpt, Theta Chatbot focuses on industry-specific content. For example, only hockey related information from NJ Devils.
Q: Are there any plans to expand “Bot Stevens” with features like predictive analytics? Do you want to get more skilled?
A: Currently, a major area of development is expanding interactive RAG chatbots to become proactive agent AI with a variety of scalable actions. For example, you can interface with customer service and CRM systems to automatically create support tickets to improve user satisfaction. An AI agent also creates “fantasy sports/esports teams” that can compete against users to increase the steps towards fan engagement and talent identification, player evaluation, and predictive analysis of opponents.
Q: What inspired you to attract Theta Lab to partner with the New Jersey Devils? What customization options are available to other teams and brands?
A: Theta Edgecloud has won emergency with professional sports teams starting with the Las Vegas Knights and the esports team, including the FlyQuest and Evil Geniuses. The user interface is fully customizable and branded to integrate into each team’s website, mobile app, Discord and other social channels. More importantly, the real-time data sources and APIs that feed into EdgeCloud’s RAG database are customized to domain-specific information such as NHL Hockey, Valorant, League of Legends, and eSports games, such as Team/League-specific information.
You might like it too: Multibank Group is rumoured to be planning to release MBG tokens
Q&A with Mark Cianpa, Vice President of Content at NJ Devils
Q: How will the team promote “Bot Stevens” to promote fan engagement and adoption?
A: To promote “Bot Stevens,” the Devils will integrate promotions across multiple platforms to promote fan interaction, the ability of “Bot Stevens” and raise public awareness.
Q: What kind of curated content does a chatbot offer beyond statistics and schedules?
A: In addition to providing statistics and schedules, Bott Stevens offers curated content to enrich the fan experience, including historical highlights and information about events and venues. “Bot Stevens” answers questions about memorable moments from the team’s history, provides details on upcoming events at the Devils and concert spaces, theme nights, and more, as well as comprehensive details on the Prudential Centre amenities. These facilities include accessibility services, bag policies, concession locations, and ATM locations throughout the venue.
Q: How do you measure chatbot success in enhancing fan engagement and sharing relevant accurate statistics/data?
A: Measure success through user engagement metrics, accuracy, fan feedback and reduced support load. It will monitor the number of interactions, session duration, and repetitive usage, track the accuracy of the information provided, and ensure that the shared statistics and data are correct and up to date. It also collects user feedback through research to understand satisfaction levels and areas of improvement, find ways to measure the reduction in inquiries handled by human support, and show the efficiency of chatbots in addressing common questions.
Q: With AI fatigue increasing among customers, how do you make “Bot Stevens” different from other AI agents that clients already meet “digitally”?
A: Through continuous learning, human-like conversation tones, personalized interactions, and creating a chatbot ecosystem. Implement machine learning algorithms that adapt to fan feedback and evolving interests, ensuring that chatbots continue to be relevant and attractive, and design chatbots’ communication styles to be friendly and relevant, improving the user experience. Leverage AI to adjust responses based on individual fan preferences and behavior, making each interaction unique. Eventually, “Bot Stevens” will be available in the app, tied to other important aspects such as concession guides, wayfinding and other ideas.
