Karatux: one school operating system, with an AI teaching assistant built for teachers
Karatux is an EdTech company serving school groups in Nigeria, Ghana and Kenya. Morphlix built a multi-tenant school operating system covering academics, admissions, finance, communication and curriculum, with an AI teaching assistant on Amazon Bedrock AgentCore. It has been in production since 18 September 2026, after an Amazon Web Services (AWS)-funded proof of concept from 2 to 7 August 2026.
- Education
- K–12
- Generative & Agentic AI on AWS
- Timeline
- Proof of concept 2–7 August 2026
- Team
- in production since 18 September 2026 · 6 engineers
- Industry
- Education
Results
~60%
less lesson-prep time
About 90 minutes per subject per week fell to about 35, measured with teachers during rollout.
88%
first-time accuracy on curriculum questions
Measured against approved reference answers at launch, and improving after a retrieval-tuning cycle.
82%
of AI recommendations accepted by teachers
The challenge
Separate systems, uneven answers
Schools ran admissions, academics, finance, communication and curriculum in separate systems. As school groups grew across campuses and countries, administrative work and uneven curriculum quality grew with them.
Disconnected operations
Admissions, academics, finance, communication and curriculum each lived in their own system.
Growth multiplied the work
Every new campus and country added administration and made curriculum quality harder to keep even.
Lesson preparation by hand
Teachers spent about 90 minutes per subject each week preparing lessons.
Answers depended on who you asked
The answer to a curriculum question depended on which person you asked.
School groups needed one platform, and an assistant that answered from the approved curriculum.
The solution
One multi-tenant school operating system
A single platform for every school in a group, with each school's data logically isolated from other schools.
Academics and curriculum
Academics, curriculum and documents in one place, with each school's own curriculum version.
Admissions
Admissions run on the same platform as the rest of school operations.
Finance and fees
Finance and fees, with a dedicated finance ledger.
Communication and attendance
Communication with families, and attendance tracking.
Assessment and results
Assessment and results, with anything touching student assessment reviewed by a teacher.
The AI layer
An AI teaching assistant teachers actually use
The assistant answers curriculum and school-policy questions from the approved curriculum, drafts lesson plans for a subject and grade, and flags content touching student assessment for a teacher to review before it reaches a parent or student.
AgentCore and Claude
It runs on Amazon Bedrock AgentCore with Anthropic Claude models on Amazon Bedrock, with an isolated session per teacher.
Retrieval per school
Curriculum content comes from Amazon OpenSearch, filtered by subject, grade and campus, so each school gets its own curriculum version.
Defined tools only
It acts through defined tools: retrieve curriculum, draft a lesson plan, answer a policy question, or flag content for teacher review.
Guardrails and isolation
Amazon Bedrock Guardrails keep content education-appropriate and grounded in approved material, and each school's data is logically isolated from other schools.
Outcome
Measured with teachers
About 60% less lesson-prep time (≈90 → ≈35 minutes per subject per week), measured with teachers during rollout., 88% first-time accuracy on curriculum questions against approved reference answers, improving after a retrieval-tuning cycle. and 82% of AI recommendations accepted by teachers.
Architecture
Multi-tenant on AWS
Edge
Amazon Route 53, Amazon CloudFront and AWS WAF in front of an Application Load Balancer, with the web front end on AWS Amplify and Amazon S3.
Application
Amazon ECS on AWS Fargate, with AWS Step Functions, AWS Lambda and event-driven messaging through Amazon SNS, Amazon SQS and Amazon EventBridge.
Data
Amazon Aurora Serverless with Amazon RDS Proxy for school operations, and Amazon RDS for PostgreSQL for the finance ledger.
AI
Amazon Bedrock and Amazon Bedrock AgentCore, with curriculum retrieval from Amazon OpenSearch Service.
Operations
AWS Secrets Manager, AWS X-Ray and Amazon CloudWatch.
Technology stack
AWS services behind Karatux
Edge and delivery
- Amazon Route 53
- Amazon CloudFront
- AWS WAF
- Application Load Balancer
- AWS Amplify
- Amazon S3
Compute and workflow
- Amazon ECS on AWS Fargate
- AWS Step Functions
- AWS Lambda
Data
- Amazon Aurora Serverless
- Amazon RDS Proxy
- Amazon RDS for PostgreSQL
- Amazon OpenSearch Service
AI
- Amazon Bedrock
- Amazon Bedrock AgentCore
Messaging
- Amazon SNS
- Amazon SQS
- Amazon EventBridge
Operations
- AWS Secrets Manager
- AWS X-Ray
- Amazon CloudWatch
More from the library.
All case studiesAI · Productivity
Seshira: AI-native project management
An engineering assistant on Amazon Bedrock AgentCore that plans sprints, turns PRDs into tasks and flags blockers, with about 80% less planning effort.
Fintech & Payments
Governing the cloud behind a payments platform
Audit trail, threat detection, alerting and reviewed Terraform changes for a payments platform heading to production.
Work with the studio
Ready to future-proof
your infrastructure?
Tell us what you are building and where the current stack is holding you back. We will come back with an architecture and a plan.