OOTB Education
Not specified
Education Technology (EdTech)
Amazon Bedrock, AWS Lambda, AWS Step Functions, Amazon S3, Amazon DynamoDB, Amazon API Gateway, AWS IAM
OOTB Education
Not specified
Education Technology (EdTech)
Amazon Bedrock, AWS Lambda, AWS Step Functions, Amazon S3, Amazon DynamoDB, Amazon API Gateway, AWS IAM
OOTB Education is an education technology company creating tailored math and language learning content for children. Their manual content creation process was high effort and difficult to scale across diverse learning objectives while maintaining rigorous pedagogical standards. Avahi designed and delivered a multi-agent Al workflow on AWS to automate content generation, embed automated quality validation, and keep human teachers in control of final approval. In three weeks, OOTB gained a scalable, safety-conscious content pipeline benchmarked for competitive performance against Google Gemini for Education.
OOTB Education is an edtech pioneer focused on technology-enabled learning solutions for children. The company produces high-quality, personalized educational content for math and language learning, ensuring materials align to learning objectives and meet strict standards appropriate for child education.
OOTB Education needed to generate customized educational content that could adapt to different age groups, learning goals, and subject requirements while maintaining consistent quality. Their existing manual creation process was time-intensive and constrained by available educator capacity, making it difficult to scale content production without compromising standards.
Beyond speed, OOTB required a workflow that could understand learning objectives, produce appropriate materials, validate quality against defined rubrics, and then route outputs to teachers for review and refinement before anything reached students. They also needed the solution to be competitive, specifically benchmarked against Google Gemini for Education. If these requirements were not addressed, OOTB risked slow content velocity, inconsistent quality at scale, and reduced ability to compete in a fast-moving Al-enabled education market.
AWS provided the building blocks to orchestrate complex, multi-step Al workflows with enterprise-grade security and scalability. Using Amazon Bedrock, OOTB could leverage multiple foundation models for generation and evaluation while avoiding bespoke model hosting overhead.
AWS serverless services enabled rapid delivery and elastic scaling. This allowed OOTB to automate content generation and validation workflows, store and track artifacts reliably, and expose standardized system endpoints for integration into product experiences.
OOTB Education selected Avahi for deep expertise designing production-ready, multi-agent Al systems on AWS that balance automation with governance and human oversight. Avahi brought practical experience implementing model routing, evaluation patterns (including LLM-as-ajudge), and workflow orchestration that aligns with real-world quality constraints.
Avahi also delivered an architecture designed for competitive benchmarking, enabling OOTB to compare capability against Google Gemini for Education while keeping the solution extensible for MVP development.
Avahi built a multi-agent Al workflow on AWS to automate educational content generation while maintaining teacher-in-the-loop governance. The workflow was orchestrated using AWS Step Functions, coordinating a multi-step pipeline that interprets learning objectives, generates candidate content, evaluates quality, and prepares refined outputs for human review. AWS Lambda executed discrete agent functions across each stage, supporting modularity and fast iteration.
The system integrated multiple foundation models through Amazon Bedrock, including Amazon Nova for specialized tasks. An intelligent routing layer selected the most appropriate model pathways depending on whether the content target was math or language learning, improving relevance and output quality across domains.
To enforce standards, Avahi implemented an LLM-as-a-judge quality validation framework. Generated content was automatically assessed against predefined rubrics and quality criteria before it reached educators, reducing review burden and helping ensure outputs remained aligned with pedagogical expectations and child safety considerations.
Content templates and generated artifacts were stored in Amazon S3, creating a durable sourcе of truth for both inputs and outputs. Metadata and workflow state were tracked in Amazon DynamoDB, enabling traceability across content versions, evaluation outcomes, and human review status. The solution exposed RESTful endpoints via Amazon API Gateway for system interaction and integration into downstream experiences. Access and permissions were managed with AWS IAM to enforce least-privilege controls across services and workflows.
In only three weeks, OOTB Education gained an automated, scalable content generation pipeline that preserved educational rigor and kept teachers in control of what students ultimately see. The solution delivered a repeatable workflow that generates age-appropriate content, validates quality automatically, and routes refined outputs to human educators for final review, enabling higher throughput without sacrificing standards. The system also established a competitive baseline benchmarked against Google Gemini for Education and provided a foundation for minimum viable product development.
Let’s explore your high-impact AI opportunities together in a complimentary session
Let’s explore your high-impact AI opportunities together in a complimentary session
Let’s explore your high-impact AI opportunities together in a complimentary half-day session
