What was the impact of the AI solution?

Up to 87.5% infrastructure cost reduction on AI pipelines. System uptime of 99%. 10.5% AI accuracy gain, with F1 on the NER workflow improving from 0.80 to 0.886. Token costs down 55% via intelligent model routing. ~99% improvement in issue resolution time, from 30 hours to minutes.

Source

Evidence: Medium50/100

Mediacorp & SoftwareOne Success Story - AWS

Use case typeIT operationsUpdated Jul 10, 2026

Mediacorp needed accurate, cost-efficient automation for large-scale generative AI and multimodal metadata enrichment across millions of media assets. It also wanted better visibility into AI model performance, token cost, accuracy, and incident debugging. Mediacorp worked with SoftwareOne to build an observability layer for AI workloads on AWS, extending Amazon CloudWatch to track model accuracy, token consumption, latency, and cost. The solution used AWS Lambda, Amazon ECS, and Amazon Bedrock model selection and routing using CloudWatch metrics.

Organization: Mediacorp
Industry: Tech & Comms
Location: Singapore
Published: July 2026

Reported outcomes

+99%

issue resolution time improvementTime & speed

−87.5%infrastructure cost reduction99%system uptime0.8-0.9%AI accuracy gain55%token costs reduction

Strategic outcomes

Cost efficiencyImproved AI workload observabilityOther strategic outcomeEnabled intelligent model routing

Catalog median for time & speed deployments: +60% across 143 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Infrastructure cost reduction: 87.5% decrease

AWS Solutions Case StudyJul 10, 2026Customer storyExplicit claimMedium evidence strength

up to 87.5% infrastructure cost reduction on AI pipelines

Normalized claim

System uptime: 99%

AWS Solutions Case StudyJul 10, 2026Customer storyExplicit claimMedium evidence strength

system uptime of 99%

Normalized claim

AI accuracy gain: 0.8-0.9% increase

AWS Solutions Case StudyJul 10, 2026Customer storyExplicit claimMedium evidence strength

10.5% AI accuracy gain (F1 on NER workflow improved from 0.80 to 0.886)

Normalized claim

Token costs reduction: 55%

AWS Solutions Case StudyJul 10, 2026Customer storyExplicit claimMedium evidence strength

token costs down 55% via intelligent model routing

Normalized claim

Issue resolution time improvement: 99% increase

AWS Solutions Case StudyJul 10, 2026Customer storyInferred claimMedium evidence strength

~99% improvement in issue resolution time (30 hours to minutes)

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Mediacorp
Provider: AWS
Maturity: Unknown
Linked source: AWS Solutions Case Study

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 3

1IT operations
2Workflow orchestration
3Observability

Mediacorp needed accurate, cost-efficient automation for large-scale generative AI and multimodal metadata enrichment across millions of media assets.
Mediacorp wanted better visibility into AI model performance, token cost, accuracy, and incident debugging.

Mediacorp worked with SoftwareOne to build an observability layer for AI workloads on AWS.
It extended Amazon CloudWatch to track model accuracy, token consumption, latency, and cost.
The solution used AWS Lambda, Amazon ECS, and Amazon Bedrock model selection and routing using CloudWatch metrics.

Technologies

Amazon CloudWatch AWS Lambda Amazon ECS Amazon Bedrock

Up to 87.5% infrastructure cost reduction on AI pipelines.
System uptime of 99%.
10.5% AI accuracy gain, with F1 on the NER workflow improving from 0.80 to 0.886.
Token costs down 55% via intelligent model routing.
~99% improvement in issue resolution time, from 30 hours to minutes.

Architecture

An observability layer for AI workloads on AWS extended Amazon CloudWatch to monitor model accuracy, token consumption, latency, and cost, while AWS Lambda and Amazon ECS supported the workflow and Amazon Bedrock model selection/routing was driven by CloudWatch metrics.

Implementation partners1

SoftwareOne

Sources & evidence1

Evidence: Medium50/100Evidence strength