This note summarizes what the public GMA case study is based on.
What is public
The current public material supports these points:
- the long-to-short video automation workflow is launched and live in production inside GMA Network's own AWS environment
- the deployment is a Customer-Deployed AWS solution operated by GMA, with Pixel ML as the AWS Partner
- the system uses Amazon Bedrock, Amazon SageMaker, Amazon EKS, EC2 G6e GPU, S3, DynamoDB, Lambda, and CloudWatch as its core AWS services
- editorial team rollout is in active progress, with the system processing real workloads
- the workflow is a multi-agent (Agentic AI) pipeline with human-in-the-loop editorial review checkpoints
- production status (not pilot, not proof of concept) is confirmed by AWS ACE-registered launched-stage opportunity records and customer-approved deployment artifacts
What the wording is based on
The case study is based on customer-approved delivery material, AWS ACE opportunity records (launched stage), and Pixel ML operational deployment artifacts for the production workflow.
What stays private
These details are not on the public site and remain available only to AWS for validation purposes:
- detailed throughput and clip-volume metrics
- internal cost economics and unit-economics analysis
- customer-specific architecture diagrams beyond the high-level summary
- full agent-by-agent IAM and resource configuration
- future-phase scope and roadmap material
Why this case matters
GMA is one of the clearest examples on the site of an Agentic AI Applications workflow running in production inside a customer's own AWS environment, with multi-agent orchestration on Amazon EKS, Foundation Model integration through Amazon Bedrock, custom Tagalog-aware models on Amazon SageMaker, and full AWS-native security and observability.