Business Problem

• While AI has reduced the story creation process from 70 to 2 days, allowing a book to be completed in a day with manual selections, Ojje aims to streamline this process further to bring it down to just a few hours.
• GCP’s infrastructure previously posed challenges to rapid content production, but Ojje has since shifted to AWS for better performance, scalability, and cost efficiency.
• Ojje is now evaluating AWS to ensure it meets operational and financial objectives. A proof of concept is needed to demonstrate faster book creation and validate AWS’s potential.

Overview

Ojje is a pioneering learning product that enhances children’s reading skills through self-guided interactive books. Its mission is to become the leading self-driven interactive book publisher globally and address the decline in reading skills globally.

Solution

GoML became a vital partner for Ojje by delivering comprehensive solutions that addressed their key challenges.

‍

GoML implemented Amazon Bedrock, utilizing an LLM for efficient story generation. This allowed Ojje to create exciting stories in just a few hours. The books are finished within a day by manually choosing the best options, making them suitable for different age groups.

‍

Using MidJourney, GoML helped Ojje create high-quality, custom illustrations that perfectly match the stories. This process ensures the storybooks are visually attractive and completed quickly without long waits.

‍

GoML deployed Ojje’s application on Amazon EC2 instances, providing the scalable computing resources necessary to handle varying loads and ensuring high availability. This infrastructure supports the rapid processing and delivery of interactive books.

‍

With AWS SQS, GoML helped make the process more efficient by organizing and managing the tasks of creating stories and images. This ensured that everything worked smoothly and tasks were completed promptly.

‍

GoML utilized MongoDB to store comprehensive story details, including metadata and prompts. This flexible data management solution allowed Ojje to organize and retrieve story-related information efficiently.

‍

GoML used AWS S3 to store generated images, providing a scalable and durable storage solution. This ensured that all illustrations were safely stored and easily accessible, allowing the quick creation and release of storybooks.

Architecture

• API Gateway
  Manages API requests and directs traffic to backend services.

• AWS IAM
  Controls user permissions and secures access to AWS resources.

• LLM on Amazon Bedrock
  Powers scalable and flexible story generation with an LLM.

• Story Writer Service
  Generates and refines story content using an LLM on Bedrock.

• Story Image Generator
  Creates visual content aligned with story prompts using MidJourney.

• MidJourney
  Provides high-quality artistic and thematic storybook images.

• AWS Lambda
  Executes serverless code in response to story and image processing events.

• AWS CloudWatch
  Monitors and logs AWS resources and application performance.

• AWS EC2
  Manages Docker container images for application deployment.

• Private Network
  Ensures secure communication between AWS resources.