Education & Learning

Building an AI-Ready Platform for Classroom Learning

Project Overview

Since December 2023, NUS Technology has worked with Baamboozle to extend and strengthen its game-based learning platform as customer demand and product requirements have continued to grow.

Baamboozle serves a global market of teachers and students, helping educators turn classroom review sessions into interactive, multiplayer games. Teachers can create quiz-style activities, display them on a shared television or projector, and invite students to participate through a classroom PIN without requiring them to install an application.

As adoption and product ambitions increased, Baamboozle’s existing development capacity was no longer sufficient to deliver new functionality at the required pace. NUS joined the project to expand the engineering team, continue development of the established platform, resolve operational issues, and support the development of a new generation of AI-assisted authoring tools.

  • Overlapping squares

    Name

    Baamboozle / Ronan Casey

  • Overlapping squares

    Services

    Stripe, AWS S3, CloudFront, Google Cloud Translate, OpenAI, Google Gemini, OpenRouter, Langfuse, Giphy, Bing Image Search, Meilisearch / Elasticsearch, Sentry

  • Overlapping squares

    Backend

    Laravel, PHP

  • Overlapping squares

    Frontend

    Next.js, React, TypeScript

  • Overlapping squares

    Infrastructure

    AWS

  • Overlapping squares

    Industry

    EdTech / Game-Based Learning

  • Overlapping squares

    Market

    Global

Overview Image client logo
Overlapping squaresTHE CHALLENGES

Scaling a Growing Product Ecosystem

Baamboozle had evolved beyond a standalone classroom game application. Its Laravel platform managed teacher accounts, educational games, subscriptions, content discovery, administration, advertising, authentication, media services, email communications, and numerous external integrations. At the same time, a newer Next.js application was being developed to support Baamboozle Studio, AI-assisted content creation, and emerging experiences such as the Baamboard interactive game board.

Supporting both environments created several interconnected challenges.

Complex Subscription Operations: Baamboozle needed to reliably manage complex subscription scenarios such as pauses, resumptions, trial extensions, cancellations, expirations, and resubscriptions while keeping Stripe, account access, expiration dates, and admin data consistent.

An Outdated Search Foundation: Baamboozle’s growing content library relied on an outdated Elasticsearch driver, requiring a new search foundation that could support educational content structures such as stemming, camelCase hashtags, nested questions, tags, and separate environment indexes.

Unreliable External Content Imports: Teachers needed to import existing content from platforms such as Kahoot and Quizlet, but differences in data structures, parsing errors, edge cases, and inappropriate content made these workflows unreliable and often required manual correction.

Continuous Platform Modernization: Baamboozle needed to upgrade Node.js, React, and Next.js while introducing multi-provider AI functionality, streaming, caching, and observability without causing regressions or disrupting the classroom experience existing users relied on.

Challenge Image client logo

Stabilizing Core Operations

NUS strengthened subscription workflows across Stripe, Laravel Cashier, the application database, and Baamboozle’s administration tools. The team resolved edge cases involving paused and resumed subscriptions, trial extensions, cancellations, free-plan expiry dates, and customers resubscribing while an existing plan remained paused.

The team also improved third-party integrations, including handling Giphy rate limits more gracefully, and introduced feature flags so selected functionality could be activated or disabled safely in production. Automated browser testing with Laravel Dusk was expanded to reduce regressions as the platform continued to evolve.

Modernizing Search and Content Workflows

NUS replaced Baamboozle’s outdated Elasticsearch driver and rebuilt search around the structure of its educational content. The new implementation supports English stemming, camelCase hashtags, nested questions, tag-based discovery, advanced search, and separate indexes for local, staging, and production environments.

Search and administrative queries were also optimized, while import workflows for Kahoot and Quizlet were strengthened with source-specific parsing, error handling, and profanity filtering. These improvements made it easier for teachers to reuse existing material and for Baamboozle’s team to manage a growing content library.

Supporting Continuous Platform Development

NUS supported major upgrades across Node.js, React, and Next.js, resolving styling, responsive design, and end-to-end regressions after each upgrade. This allowed Baamboozle to continue modernizing its technology stack without interrupting the development of new functionality.

Building AI-Assisted Authoring

For Baamboozle Studio, NUS developed an extensible AI Toolbar that allows teachers to translate games and questions directly within the editor. Teachers can review, edit, approve, or reject individual translations before applying them, keeping control of the final educational content.

A resilient batch translation pipeline processes up to 100 items per request, caches individual results, isolates failed batches, and applies timeouts to prevent stalled jobs. The surrounding architecture was designed to support future tools such as content improvement, simplification, and hint generation.

NUS also introduced a multi-provider AI routing layer supporting models from OpenAI, Google Gemini, and Claude through OpenRouter, Kimi, and other providers available through the Vercel AI SDK. Model-specific settings can be changed without modifying the surrounding product logic.

Redis-based caching supports both streaming and non-streaming AI responses and follows a fail-open design, allowing AI features to continue operating if the cache becomes unavailable. Langfuse tracing and LLM-as-judge scoring provide Baamboozle with a foundation for comparing model outputs and improving AI quality over time.

Information Banner client logo
Overlapping squaresTHE RESULT

A Stronger Foundation for Continued Growth

Since NUS joined the project, Baamboozle has strengthened its development capacity, platform reliability, performance, and content workflows, delivering measurable improvements across four key areas:

  • Green tick circle

    Increased Development Capacity: Expanded engineering support gave Baamboozle approximately 35% more effective development capacity, helping the team maintain the existing platform while advancing Baamboozle Studio and AI-assisted authoring.

  • Green tick circle

    Improved Subscription Reliability: Enhancements to subscription lifecycle handling made pause, resume, cancellation, trial extension, and resubscription scenarios more dependable, resulting in an estimated 30% reduction in recurring billing and account-access issues.

  • Green tick circle

    Enhanced Loading Times: Search migration and query optimization improved performance across key administrative and team-management pages, delivering approximately 25% faster loading times while creating a more maintainable search foundation.

  • Green tick circle

    Faster Content Preparation: Improved Kahoot and Quizlet imports, combined with batch AI translation, cut the time needed to prepare or migrate a typical classroom game by an estimated 40%, helping teachers reuse and process content more efficiently.

Result Image client logo
Overlapping squares

Effective development capacity increased by approximately 35%

Overlapping squares

Billing and account-access issues reduced by 30%

Overlapping squares

25% faster loading times

Overlapping squares

40% reduction in game preparation and migration time

Circle check

Conclusion

From Classroom Games to an Extensible Learning Platform

Baamboozle’s continued growth has expanded the product far beyond its original classroom game experience. Subscriptions, educational content, search, external imports, advertising, media services, AI providers, analytics, and teacher-facing authoring tools now operate as components of one connected global platform hosted on AWS.

Since joining the project in December 2023, NUS has helped Baamboozle strengthen these connections while continuing to build new functionality. By resolving operational edge cases, modernizing foundational systems, supporting major framework upgrades, and introducing a flexible AI architecture, NUS has enabled Baamboozle to respond to growing product demand without neglecting the reliability of the platform teachers already use.

Conclusion Image

Links

Download This Case Study

Get the full story in PDF format, including project details, technical insights, and measurable results you can apply to your own business.

Download backgroundDownload background

Explore Other Case Studies

Case study australian-motorcycle-ecommerce-transformation
Commerce & Marketplace

Orchestrating Complex E-Commerce Operations for an Australian Motorcycle Parts Retailer

Case study vesta-living
Property & Hospitality

Propelling a Seamless Property Management Experience: A Case Study on VestaLiving

Ready to Write Your Own
Success Story?

Let’s have a strategic conversation about your goals and how we can achieve results like these for your business.

Schedule a Strategy Session
CodeMonitorGrid with light