The Beam’s School Board Chatbot
The Arizona State University Artificial Intelligence Cloud Innovation Center, powered by Amazon Web Services (AWS), collaborated with The Beam, a media outlet under the Cronkite School’s Howard Center for Investigative Journalism, to explore how artificial intelligence can improve access to information from public school board meetings across Arizona. Many school districts publish meeting recordings on YouTube as part of transparency efforts, but these recordings are often several hours long and difficult for parents, journalists, and community members to navigate efficiently.
The project focused on re-developing and expanding the team’s AI-powered chatbot that allows users to ask specific questions about school board meetings directly using transcripts. The solution was designed to provide accurate, context-aware responses while reducing the time to manually review lengthy school board meetings.
The Beam plans to launch this new chatbot soon.
Problem
Parents and journalists often need to locate highly specific information within school board meetings, such as discussions about school closures, policy changes, or controversies. While there are always meeting minutes, they are typically very brief. There is currently no user-friendly way to search for this detailed information effectively.
The Beam identified the need for a chatbot experience that could help users quickly access relevant information from recent school board meetings without requiring external accounts or extensive manual searching.
Using an OpenAI product called ChatGPT, The Beam launched school board chatbots powered by meeting transcripts that were very popular. But users need ChatGPT accounts, and the free version limits the number of queries. The Beam wanted a more accessible option.
Student Spotlight
Role: Full Stack Developer | Connect with Shawn on LinkedIn
Developing a system for the School Board Chatbot project has been an amazing opportunity to learn and leverage new advancements in AI technology. We got to experiment with new agile development methodologies like mob coding while engineering a solution that will help support over 70+ school districts across Arizona. Working very closely with The Beam team while testing this new method of development has been a wonderful experience and has enabled us to receive immediate feedback, maintain scope clarity, and accelerate through collaboration. One of the biggest roadblocks faced was YouTube’s API limitations and working around this in a graceful fashion while still providing a solution that best addressed their needs. Throughout this project, the collaboration between the CIC and Cronkite teams is ultimately what made this project so successful and I’m glad to have been a part of it.
Role: Full-Stack Developer | Connect with Shakthi on LinkedIn
The most valuable part of working on the School Board Chatbot project was expanding my experience beyond design and learning the full-stack development process through collaborative mob coding and mob elaboration sessions and developing the core application in 6 hours. As a Full-Stack Developer, I contributed to both frontend and backend development while also participating in technical documentation and architecture diagram discussions, which gave me a deeper understanding of how large-scale applications are structured and developed. Through this experience, I learned the importance of asking questions, volunteering to help with tasks, and managing parallel work streams to contribute more effectively within a fast-paced team environment. I also developed stronger confidence in asserting my ideas during technical discussions and gained a greater appreciation for collaboration, communication, and cross-functional problem-solving throughout the development process.
Role: UI/UX Designer | Connect with Jenny on LinkedIn
The most exciting part of working on The Beam’sSchool Board Chatbot was reimagining how parents, teachers, and school leaders interact with their local school board meetings, helping them find updated information around school safety and the decisions shaping their communities. As the UX designer, I was responsible for auditing the current site, defining the user flows for district selection and chat interaction, and designing the visual system in Figma while preserving The Beam's editorial brand. Throughout the project, I was able to challenge myself and learn from my peers about Mob coding, which allowed us to do a live demonstration of the application with the client and show the behind-the-scenes of the code as it was being built. I learned how much faster ideas can move from design into a working product when designers and developers collaborate in real time, and how powerful it is to let clients see the build happen rather than only the final result.
Approach
The Beam initially built a set of AI-powered chatbots on ChatGPT using custom GPTs to let users query recent school board meeting transcripts. The ASU AI CIC team helped The Beam develop its own standalone version of this tool, improving the user experience and expanding the system from 12 to 70 chatbots. Selected school district YouTube channels are monitored. Retrieved transcripts are then stored in a centralized system for chatbot retrieval and administrative access. The solution was designed to support transcript storage, retrieval, and conversational search while helping users quickly access information from recent school board meetings.
- AI-Powered Q&A: AWS Bedrock Knowledge Base uses Claude Haiku 4.5 for response generation and Amazon Titan Embed Text v2 to index and retrieve transcript content, allowing users to ask questions and receive answers with transcript citations.
- District-Specific Search: Users select a school district before interacting with the chatbot. Responses are generated using transcript content associated with the selected district's most recent board meetings.
- Monitoring and Analytics: EventBridge triggers automated YouTube channel monitoring every six hours, while Amazon DynamoDB stores transcript metadata and anonymized query logs used to identify community interests and trends.
- Frontend and Administration: A Next.js application hosted on AWS Amplify provides the public chatbot experience and administrative dashboard. Amazon Cognito secures administrator access for district management, transcript uploads, and analytics.
- Infrastructure: AWS CDK deploys a fully serverless architecture using API Gateway, Lambda, Bedrock Knowledge Base, DynamoDB, S3, and Transcribe.
Industry Impact and Problem Solving
The solution includes logging of anonymized user queries, allowing The Beam to generate editorial insights and identify trending community concerns. It explores an approach for providing accurate responses to user questions while using recent meeting transcripts as the primary source of information.
This proof of concept demonstrates how artificial intelligence can improve public access to local government and education information by making lengthy school board meetings more searchable and accessible. Overall, this helps reduce the time required to locate important information.
This was a mind boggling experience. I sat on a couch, eating my CIC-provided Chipotle bowl, and watched as the team mob coded (a word I learned through this process!) our chatbot. There was so much value in being able to provide feedback in real time, and I learned new vibe coding strategies. The team also set me up so that I can use Claude Code to troubleshoot and improve the chatbot independently going forward.
-Shelby Grossman, Professor of Practice of AI and Investigative Journalism, Howard Center for Investigative Journalism, Cronkite School
Mob coding with the AWS team was incredibly valuable for developing my AI knowledge and skills. It was a collaborative way to not only build the chatbot efficiently, but also ensure accuracy at every step of the process. That experience gave me the confidence to apply what I'd learned and build my own portfolio website using a similar vibe coding approach. Coding no longer feels as daunting to me, and despite my limited background in it, I now feel comfortable handling the basics.
-Lizzie Tomlin, Intern for The Beam and M.A. in investigative journalism student at the Cronkite School
Potential for Wider Application
While this proof of concept was designed for The Beam’s school board reporting initiatives, the solution framework has potential for growth. Future scope discussions for the project include expansion to additional school districts across Arizona.
Other possible enhancements for the future include automated identification of newsworthy or controversial moments and integration with newsroom tools and workflows.
Supporting Artifacts
| Github Link: | Click Here |
Next Steps
This project served as a proof of concept for how generative AI can support public access to local education and government information. By demonstrating the feasibility of full-transcript querying, and conversational access to school board discussions, the project established a foundation for future expansion and refinement.
Future exploration may include scaling the solution to additional school districts and integrating newsroom workflows that help journalists identify emerging topics and community concerns more efficiently.
About the ASU CIC
The ASU Artificial Intelligence Cloud Innovation Center (AI CIC), powered by AWS, is a no-cost design thinking and rapid prototyping shop dedicated to bridging the digital divide and driving innovation in the nonprofit, healthcare, education, and government sectors.
Our expert team harnesses Amazon’s pioneering approach to dive deep into high-priority pain points, meticulously define challenges, and craft strategic solutions. We collaborate with AWS solutions architects and talented student workers to develop tailored prototypes showcasing how advanced technology can tackle operational and mission-related challenges.
Discover how we use technology to drive innovation by visiting ASU AI CIC or contacting [email protected].