AWS Builder Center
Weekend Creative Challenge: BackBench AI

Weekend Creative Challenge: BackBench AI

An emotional, immersive AI storytelling web application built with Amazon Bedrock (Nova Lite), Next.js 15, and AWS Amplify that transforms late 1990s and 2000s Indian school memories into diaries, teacher remarks, school magazine articles, and Bollywood scripts.

AWS CB | DevOps Engineer

Vision & What the App Does

BackBench AI is an emotional, immersive storytelling web application built to relive Indian school memories from the late 1990s and early 2000s.
Instead of building another generic chatbot, BackBench AI transports users back into a vintage 90s classroom (Class 9B). The experience combines semi-realistic digital artwork of wooden desks, green chalkboards, and dusty sunlight with a real-time Web Audio API soundscape (monsoon rain outside, ceiling fan sub-hum, distant corridor chatter, and brass school bell chimes).

Figure 1: The BackBench AI Hero Interface featuring dusty window sunlight and ambient soundscape controls.
When a user shares a memory - such as "I forgot my homework", "My first crush", "Our PT period", or "I was always on the last bench" - BackBench AI transforms that single memory into 4 distinct creative outputs:
  1. School Diary: An emotional handwritten diary page written from the perspective of an Indian school student in 2002, complete with lined parchment textures and ink blotches.
  1. Teacher Remarks: A red-ink exercise book correction box displaying a funny, authentic teacher feedback remark (e.g. "You have great potential. If only you talked less during mathematics.") with a grade stamp and red signature.
  1. School Magazine: A newsprint article column styled like an annual school souvenir edition ("Saraswati Vidyalaya Annual Edition 2002-03"), complete with drop caps and a highlight quote box.
  1. Bollywood Scene: A dramatic screenplay script ("INT. CLASSROOM 9B - RAINY AFTERNOON") with character tags and hilarious 2000s Bollywood-style dialogues in Hinglish.



How I Built It

The application was built from scratch in one focused weekend using modern web engineering principles:
  • Frontend Framework: Next.js 15 App Router with React 19 and TypeScript.
  • Styling & Aesthetics: Custom TailwindCSS design system with warm nostalgic tokens ( #140c08  dark wood,  #1a3328  chalkboard green,  #f8f2e4  cream paper, and  #d4af37  golden sunlight).
  • Interactive Visuals: HTML5 Canvas rendering floating golden dust motes illuminated by diagonal sunlight beams coming from the classroom window.
  • Web Audio Soundscape Engine: Procedural sound generation using Web Audio API nodes ( AudioContext ,  BiquadFilter , sub-oscillators) for monsoon rain, ceiling fan blade swoosh, and dual-sine brass school bell chimes.
  • Animations: Framer Motion spring physics for smooth modal transitions and paper unfolding animations.

Figure 2: Memory input modal allowing users to pick classic 90s memories or enter custom prompt.

Figure 3: Generated School Diary.

Figure 4: Generated Red-Ink Teacher Remarks.

Figure 5: Generated School Magazine.

Figure 6: Generated Bollywood Scene.

Challenges & How I Overcame Them

During deployment to AWS Amplify, I encountered an issue where environment variables starting with  AWS_  (such as  AWS_ACCESS_KEY_ID ) were blocked because Amplify reserves the  AWS_  prefix for system runtimes. I resolved this by updating the backend service layer to support custom prefixes ( BEDROCK_AWS_ACCESS_KEY_ID  and  BEDROCK_AWS_SECRET_ACCESS_KEY ) while maintaining seamless fallback support for local development.



AWS Services Used & Architecture Overview

BackBench AI leverages the power of AWS serverless and generative AI infrastructure:

Figure 7: BackBench AI - Architecture Diagram.

Figure 8: AWS Amplify Console showing successful Next.js deployment pipeline.

AWS Services Summary

  1. AWS Amplify: Hosts the Next.js 15 SSR web application with automated GitHub CI/CD deployments, environment variable management, serverless runtime, and global SSL delivery.
  1. Amazon Bedrock: Powers the core creative story generation using Amazon Nova Lite ( amazon.nova-lite-v1:0 ). Bedrock is invoked via  @aws-sdk/client-bedrock-runtime  to produce structured JSON payloads containing the diary, teacher remarks, school magazine, and Bollywood script.



What I Learned

  1. Amazon Nova Models on Bedrock: Working with Amazon Nova Lite showed how fast and cost-effective Nova models are for structured JSON creative storytelling tasks.
  1. Combining Web Audio with Generative Text: Creating procedural Web Audio synths (rain, fan hum, brass bell) alongside AI-generated text transforms a static text response into an emotional 3D storytelling experience.
  1. AWS Amplify Next.js 15 Integration: Gained hands-on experience configuring Amplify environment variables, build settings, and secret prefix policies for Next.js 15 App Router.



Project Links



Built with ❤️ for the AWS Builder Center Weekend Creative Challenge (August 2026).
#creative-expression
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article