Bilingual AI Support Chatbot: PLTW | Artificial Intelligence Cloud Innovation Center

Overview

The Arizona State University Artificial Intelligence Cloud Innovation Center, powered by Amazon Web Services (AWS), collaborated with Project Lead The Way (PLTW) to develop a proof of concept of an intelligent AI-powered customer support chatbot that would help US PreK–12 educators (administrators and teachers) get instant answers related to PLTW curriculum implementation, product purchasing, training, assessments, payments, rostering, grants, and technical guidance. The chatbot was built using AWS Bedrock Agent and a Next.js web experience to provide fast, reliable, citation-based support at scale.

Problem

US PreK–12 educators and administrators across the United States frequently need quick answers to product, implementation, and technical questions. While PLTW supports its customers with a best-in-class Solution Center and maintains robust documentation, it can be time-consuming for educators to search for the right information, driving increased reliance on the PLTW Solution Center for tier-1 support. This also limits visibility into common question themes, satisfaction trends, and escalation drivers.

PLTW’s goal was to explore introducing a first line of support that enables self-service for tier-1 questions, improves the customer experience, and reduces call and email volume—while ensuring there is a clear escalation path when a question can’t be resolved via documentation.

Student Spotlight

Role: Cloud Developer | Connect with Pratiksha on LinkedIn
As the Developer, I created a smart AI chatbot called Jordan for PLTW that enables educators to get instant help at any hour. My work focused on three critical technical challenges - designing a scalable AWS architecture, optimizing the knowledge base integration to accurately search through support documents to deliver contextual answers without hallucination and building an intuitive interface that makes getting help as simple as asking a question. I developed the frontend from the UI design and also worked on developing important features like chatbot and admins APIs. The most rewarding aspect has been solving real-world problems that directly empower educators by transforming the frustrating experience of searching for answers into an instant, seamless interaction. By focusing relentlessly on removing barriers to enabling 24/7 self-service this project helped me understand how things can be made easier, thus allowing the educators to focus more on empowering students.

Role: Cloud Developer | Connect with Sreeram on LinkedIn
Working on the Jordan PLTW chatbot was a great experience, as I had the opportunity to contribute to both the frontend and backend development. This chatbot plays a key role in ensuring that teachers and administrators can easily access important information as part of the PLTW program. I focused on improving the accuracy of chatbot responses by integrating the AWS Bedrock Knowledge Base. Additionally, I implemented an agentic feedback system that provides administrators with conversation metrics without requiring explicit user feedback. A background agent runs a few hours after each session to analyze the conversation and identify areas for improving the chatbot from an administrative perspective. I truly enjoyed working on this project because it represented a real-world use case that benefits many teachers and administrators.

\

Role: UI/UX Designer | Connect with Omdevsinh on LinkedIn
I served as the UI/UX designer for Jordan, an AI customer-support chatbot for PLTW that helps US PreK–12 educators get fast answers on implementation, training, rostering, payments, grants, and more. Talking about my experience, I mapped user flows and produced high-fidelity Figma prototypes, defining Jordan’s friendly persona, bilingual (English/Spanish) interactions, citation and feedback patterns, and a clear escalation path when an answer isn’t found. I also designed a single-page admin dashboard to track conversation volume, satisfaction, and escalation reasons alongside a simple conversation log. The work enabled faster self-service for educators and helped reduce Solution Center workload while supporting a consistent, customer-centric experience. Working on this project helped me learn how to design for trust in AI support, balancing clarity, provenance, and seamless handoff to humans.

Approach

The CIC team built a serverless, bilingual (English/Spanish) AI support chatbot on AWS with three main flows: Educator Chat, Admin Dashboard, and Knowledge Base Enablement.

AI/RAG: Amazon Bedrock Agent grounded responses using the PLTW knowledge base; Amazon OpenSearch is referenced as a supporting component for knowledge retrieval context.
Chat & APIs: AWS Lambda powers the chatbot and admin backend services behind Amazon API Gateway, returning citation-backed answers and handling escalation logic (including queue ID generation).
Knowledge base enablement: PLTW documents are stored in Amazon S3 and used to support the chatbot’s citation-based responses and documentation linking (with special handling to avoid linking directly to PDFs).
Frontend: A Next.js web experience embeds the chatbot on PLTW’s website, delivering bilingual responses and a thumbs-up/thumbs-down feedback mechanism.
Admin & analytics: An admin dashboard secured with Amazon Cognito provides visibility into conversation logs, feedback, and metrics stored in Amazon DynamoDB.
Escalation workflow: When the chatbot cannot resolve an issue or receives negative feedback, Amazon SQS supports queue-based escalation for follow-up support.
Infrastructure: AWS CDK provisions the serverless stack, supported by a one-step deployment script to streamline setup and repeatable environments.

Industry Impact and Problem Solving

The chatbot directly addresses PLTW’s need to provide educators with fast, trustworthy answers without requiring them to navigate multiple documentation sources or wait for live support. The combination of citation-based responses, bilingual support, feedback collection, and escalation workflows improves educator self-service while maintaining a clear pathway to the Solution Center for complex cases.

Key benefits include:

For educators: immediate access to reliable answers, reduced time spent searching, and embedded resource guidance.
For the PLTW Solution Center: reduced call/email volume, improved operational efficiency, and a more consistent 24×7 customer experience.
For internal stakeholders: improved visibility into conversation volume, satisfaction trends, top question categories, escalation reasons, and conversation logs via the admin dashboard.

Collaborating with AWS and ASU AI Cloud Innovation Center was a great opportunity for us to accelerate momentum towards an initiative we’ve been exploring for some time - developing an AI Customer Support Chatbot. With customer centricity as one of our core values at PLTW, we continually look for opportunities to enhance the overall user experience while helping our educators get answers to questions quicker. The team of ASU students not only delivered a functional proof-of-concept for us, along with the architecture, but they also anticipated additional needs and created a dashboard to track metrics on interactions with the chatbot. I was continuously impressed by all their students’ efforts in this project.
Brian Greiwe, Chief Technology Officer, Project Lead The Way

Potential for Wider Application

The chatbot’s architecture is reusable for other education and public-sector organizations that need:

Tier-1 support automation with consistent messaging
Bilingual self-service experiences
Citation-based answers grounded in approved documentation
Admin visibility into usage, satisfaction, and escalation patterns

In addition, the escalation approach is extensible Workflows. PLTW noted that Amazon Connect is used for its call center, and a production implementation could include direct queue transfer as part of the escalation experience.

Next Steps

PLTW plans to leverage the work done by the ASU AI Cloud Innovation Center and further enhance the support chatbot by integrating it with Amazon Connect. This will allow customers to engage with the chatbot for quick answers to questions as well as easily be connected to a live agent should they need additional support. Future phases could expand these capabilities to support prospective customers, guiding them through the process of bringing PLTW programs to their students.

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 a wide range of operational and mission-related challenges.

Discover how we use technology to drive innovation. Visit our website at ASU AI CIC or contact us directly at [email protected].

On This Page: