Case Studies

Case studies

How we've helped customers put AI into production across financial services, media, aviation, and consumer brands.

Overview

Real deployments, named customers

From campaign personalization to customer-cloud systems, each case says what was built, where it runs, and how far it went.

Open trust centerView AWS Marketplace
13

Public customer cases

Customer work across campaign, rollout, and production.

12

Named customer references

Named customer stories with scope labels and source links.

8

Customer-cloud deployments

Cases that show customer AWS or managed delivery.

7

Rollout and production work

Cases currently in rollout or production.

Featured work

Featured customer work with clear scope labels and public references.

Generative AIInternal Product

AI Art Platform Scaled to 1.5M+ Users

QuickQR.Art demonstrated product-scale AI deployment with 1.5M+ users, proving the team can build and operate consumer-grade generative AI systems.

Proof source
Internal product data
Scope
Live product, 1.5M+ users
Deployment setting
Consumer-scale generative AI deployed on AWS infrastructure.

Public references

Case pageQuickQR.Art
Internal ProductInternal Product

AgenticFlow SaaS — Internal Product Proof

AgenticFlow runs Pixel ML's own operations as the company's SaaS product. This case is internal product proof, not evidence for Studio or Sentinel customer workflows.

Proof source
Internal product data
Scope
Production platform, internally dogfooded
Deployment setting
AgenticFlow and Agentic Video both integrate with AWS services, while enterprise deployments can run in customer-controlled AWS environments.

Public references

Case pageAgenticFlowAgenticFlow Docs
Agentic AIVinasoy

Hyper-Personalized Customer Engagement via Agentic AI at 300,000+ Scale

Pixel ML's Agentic AI customer-engagement platform delivered 300,000+ uniquely personalized 1:1 AI video greetings for Vinasoy's Tet 2025 program in Vietnam. Each consumer received a one-to-one personalized video produced by an agentic workflow on Amazon Bedrock, SageMaker, EC2 GPU, S3, and CloudFront, with Generative AI (TTS + lip-sync) used as one tool inside the agent chain.

Proof source
Customer-approved production case
Scope
Production deployment, live on AWS, 300,000+ uniquely personalized customer outputs delivered
Deployment setting
Agentic AI Applications competency category — production SaaS-on-AWS deployment using AgenticFlow on Amazon EKS, with Amazon Bedrock and Amazon SageMaker as Foundation Model inference layers, plus EC2 GPU, S3, and CloudFront.

Public references

Case pageVinasoy campaign source note
Agentic AIEnfagrow / Mead Johnson Nutrition

Hyper-Personalized 1:1 Customer Engagement via Agentic AI

Pixel ML's Agentic AI customer-engagement platform delivered 10,000+ uniquely personalized 1:1 customer interactions for Enfagrow / Mead Johnson Nutrition Vietnam. Each mother received a one-to-one AI-generated experience produced by an agentic workflow on Amazon Bedrock, SageMaker, EC2 GPU, S3, and CloudFront, with CRM signal capture on every interaction.

Proof source
Customer-approved production case
Scope
Production deployment, live on AWS, 10,000+ uniquely personalized customer outputs delivered
Deployment setting
Agentic AI Applications competency category — production SaaS-on-AWS deployment using AgenticFlow on Amazon EKS, with Amazon Bedrock and Amazon SageMaker as Foundation Model inference layers, plus EC2 GPU, S3, and CloudFront.

Public references

Case pageEnfagrow campaign source note
Agentic AICardX (SCB X Group)

Hyper-Personalized Customer Engagement via Agentic AI in Production on AWS

Pixel ML's Agentic AI customer-engagement platform is deployed in CardX's own AWS environment, in production for hyper-personalized next-best-offer marketing across 16 customer segments and 20+ rate/fee combinations under regulated risk constraints. AgenticFlow on Amazon EKS coordinates segmentation, decisioning, content variant generation, compliance validation, and reviewer agents on Bedrock, SageMaker, S3, Glue, Lambda, KMS, and CloudTrail.

Proof source
Customer-deployed AWS production case with public corroboration
Scope
Production deployment in customer's AWS environment, live since October 2025
Deployment setting
Agentic AI Applications competency category — production customer-deployed solution on AgenticFlow runtime, Amazon Bedrock, SageMaker, EKS, and full AWS-managed compute, storage, security, and analytics stack. Generative AI is used as one tool inside the agentic workflow.

Public references

Case pageCardX public article (SCB corporate site)CardX source note
GMA Network — Hyper-Personalized Editorial Engagement via Agentic AI in Customer AWS
Customer-approved production case in customer-controlled AWSProduction deployment in customer's AWS environment, live and processing editorial workloads
Agentic AI

GMA Network

GMA Network — Hyper-Personalized Editorial Engagement via Agentic AI in Customer AWS

Pixel ML's Agentic AI customer-engagement platform is deployed inside GMA Network's own AWS environment, in production for hyper-personalized editorial workflows. AgenticFlow on Amazon EKS coordinates ingest, audio (Tagalog ASR), vision, search (OpenSearch), reframe, highlights, and orchestrator agents on Bedrock, SageMaker, EC2 G6e GPU, S3, DynamoDB, Lambda, EventBridge, and CloudWatch — Generative AI is used as one tool inside the agentic workflow.

Proof: Customer-approved production case in customer-controlled AWSScope: Production deployment in customer's AWS environment, live and processing editorial workloads

Deployment setting: Agentic AI Applications competency category — production customer-deployed solution on Amazon Bedrock, SageMaker, OpenSearch Service, EKS, EC2 G6e GPU, S3, DynamoDB, Lambda, EventBridge, Cognito, and CloudWatch.

CardX (SCB X Group) — Hyper-Personalized Customer Engagement via Agentic AI in Production on AWS
Customer-deployed AWS production case with public corroborationProduction deployment in customer's AWS environment, live since October 2025
Agentic AI

CardX

CardX (SCB X Group) — Hyper-Personalized Customer Engagement via Agentic AI in Production on AWS

Pixel ML's Agentic AI customer-engagement platform is deployed in CardX's own AWS environment, in production for hyper-personalized next-best-offer marketing across 16 customer segments and 20+ rate/fee combinations under regulated risk constraints. AgenticFlow on Amazon EKS coordinates segmentation, decisioning, content variant generation, compliance validation, and reviewer agents on Bedrock, SageMaker, S3, Glue, Lambda, KMS, and CloudTrail.

Proof: Customer-deployed AWS production case with public corroborationScope: Production deployment in customer's AWS environment, live since October 2025

Deployment setting: Agentic AI Applications competency category — production customer-deployed solution on AgenticFlow runtime, Amazon Bedrock, SageMaker, EKS, and full AWS-managed compute, storage, security, and analytics stack. Generative AI is used as one tool inside the agentic workflow.

APAC Aviation Group — Investigation Workflow Deployed in Customer AWS
Anonymized customer caseCustomer-hosted phased rollout
Agentic AI

APAC Aviation Group

APAC Aviation Group — Investigation Workflow Deployed in Customer AWS

A major APAC aviation group deployed customer-hosted search, summarization, and investigation workflows in its own AWS environment.

Proof: Anonymized customer caseScope: Customer-hosted phased rollout

Deployment setting: Customer-controlled AWS deployment for high-trust investigation and review workflows. Submitted separately under aviation engagement governance, not part of Pixel ML's primary AWS AI Competency case set.

Cake by VPBank — Hyper-Personalized Customer Engagement via Agentic AI on AWS
Public-named customer with public corroborationProduction deployment, live in the Cake mobile banking app on AWS
Agentic AI

Cake by VPBank

Cake by VPBank — Hyper-Personalized Customer Engagement via Agentic AI on AWS

Pixel ML's Agentic AI customer-engagement platform delivers hyper-personalized in-app experiences inside Cake by VPBank's mobile banking app, in production on AWS. An agentic workflow tailors a unique AI-generated artistic QR code to every individual customer — combining per-user style decisioning, brand-safe generation, and 100% functional-integrity validation.

Proof: Public-named customer with public corroborationScope: Production deployment, live in the Cake mobile banking app on AWS

Deployment setting: Agentic AI Applications competency category — production SaaS-on-AWS deployment using AgenticFlow on Amazon EKS, with Amazon Bedrock and Amazon SageMaker as Foundation Model inference layers, plus EC2 GPU, S3, CloudFront, and AWS KMS.

Cho Tot Brought AI Image Generation to CSE Job Fair 2025 with Pixel ML
Customer event postEvent activation delivered
Generative AI

Cho Tot

Cho Tot Brought AI Image Generation to CSE Job Fair 2025 with Pixel ML

Cho Tot partnered with Pixel ML for a Generative AI interactive image generation experience at CSE Job Fair 2025.

Proof: Customer event postScope: Event activation delivered
Enfagrow — Hyper-Personalized 1:1 Customer Engagement via Agentic AI on AWS
Customer-approved production caseProduction deployment, live on AWS, 10,000+ uniquely personalized customer outputs delivered
Agentic AI

Enfagrow

Enfagrow — Hyper-Personalized 1:1 Customer Engagement via Agentic AI on AWS

Pixel ML's Agentic AI customer-engagement platform delivered 10,000+ uniquely personalized 1:1 customer interactions for Enfagrow / Mead Johnson Nutrition Vietnam. Each mother received a one-to-one AI-generated experience produced by an agentic workflow on Amazon Bedrock, SageMaker, EC2 GPU, S3, and CloudFront, with CRM signal capture on every interaction.

Proof: Customer-approved production caseScope: Production deployment, live on AWS, 10,000+ uniquely personalized customer outputs delivered

Deployment setting: Agentic AI Applications competency category — production SaaS-on-AWS deployment using AgenticFlow on Amazon EKS, with Amazon Bedrock and Amazon SageMaker as Foundation Model inference layers, plus EC2 GPU, S3, and CloudFront.

Vinasoy — Hyper-Personalized Customer Engagement via Agentic AI at 300,000+ Scale on AWS
Customer-approved production caseProduction deployment, live on AWS, 300,000+ uniquely personalized customer outputs delivered
Agentic AI

Vinasoy

Vinasoy — Hyper-Personalized Customer Engagement via Agentic AI at 300,000+ Scale on AWS

Pixel ML's Agentic AI customer-engagement platform delivered 300,000+ uniquely personalized 1:1 AI video greetings for Vinasoy's Tet 2025 program in Vietnam. Each consumer received a one-to-one personalized video produced by an agentic workflow on Amazon Bedrock, SageMaker, EC2 GPU, S3, and CloudFront, with Generative AI (TTS + lip-sync) used as one tool inside the agent chain.

Proof: Customer-approved production caseScope: Production deployment, live on AWS, 300,000+ uniquely personalized customer outputs delivered

Deployment setting: Agentic AI Applications competency category — production SaaS-on-AWS deployment using AgenticFlow on Amazon EKS, with Amazon Bedrock and Amazon SageMaker as Foundation Model inference layers, plus EC2 GPU, S3, and CloudFront.

Intel Featured AgenticFlow in Agentic and Multimodal AI Ecosystem Work
Public ecosystem coverageCollaboration publicly documented
Agentic AI

Intel

Intel Featured AgenticFlow in Agentic and Multimodal AI Ecosystem Work

Intel's public developer ecosystem coverage featured Pixel ML and AgenticFlow in community work around agentic and multimodal AI.

Proof: Public ecosystem coverageScope: Collaboration publicly documented
MMA Impact Vietnam 2024: AI Experience Activation for Live Audiences
Public event referenceEvent activation delivered
Generative AI

MMA

MMA Impact Vietnam 2024: AI Experience Activation for Live Audiences

Pixel ML delivered Generative AI interactive experiences at MMA Impact Vietnam 2024, bringing generative AI directly to event audiences.

Proof: Public event referenceScope: Event activation delivered
AgenticFlow SaaS — Internal Product Proof for Pixel ML
Internal product dataProduction platform, internally dogfooded
Internal Product

Pixel ML

AgenticFlow SaaS — Internal Product Proof for Pixel ML

AgenticFlow runs Pixel ML's own operations as the company's SaaS product. This case is internal product proof, not evidence for Studio or Sentinel customer workflows.

Proof: Internal product dataScope: Production platform, internally dogfooded

Deployment setting: AgenticFlow and Agentic Video both integrate with AWS services, while enterprise deployments can run in customer-controlled AWS environments.