At a glance
What this case covers
What shipped, where it runs, and the sources behind this case.
Source
Public article + customer-approved material
Scope
Production in customer environment
Basis
Based on CardX public coverage plus customer-approved implementation context.
Deployment setting
Production workflow deployed in CardX's own AWS environment.
Deployment setting
Offer-optimization workflow deployed inside the customer-controlled environment.
Not public
Detailed KPIs, offer economics, and architecture remain private.
Public references
Want more context?
See how we label scope, deployment, and source links across cases.
Overview
CardX's AI COE has publicly described its hyper-personalization work as part of its effort to reimagine customer experience in financial services through what it calls “AI with Soul.”
Pixel ML supported this direction through AgenticFlow Enterprise by building the modeling and optimization layer used to improve how offers were designed, tested, and run. The system operated in the customer environment for production use, while sensitive commercial data stayed private.
Challenges
- One-size-fits-most offers: Traditional targeting left value on the table across customer segments.
- Slow experimentation cycles: Testing new offer combinations required too much manual coordination.
- Regulated operating environment: Data handling and review standards had to fit financial-services constraints.
- Operational complexity: The customer needed outputs that could feed campaign execution reliably, not just model experiments.
Solution
Segment-Level Offer Optimization
Pixel ML built models and optimization workflows through AgenticFlow Enterprise to help CardX evaluate rate-and-fee combinations across customer segments and produce campaign-ready outputs for activation.
Customer-Environment Deployment
The workflow ran via AgenticFlow Enterprise in the customer environment, not shared SaaS. That gave the team direct control over data and operations.
Test-and-Learn Execution
The solution was designed to support faster experimentation, more structured testing, and more repeatable decisioning under business and risk constraints.
Results
- Hyper-personalization moved from concept into production use
- Offer experimentation became more structured and repeatable
- Customer experience innovation was linked to operational discipline and regulatory alignment
Why It Matters
This case shows Pixel ML delivering Agentic AI inside a regulated financial setting where relevance and operational control both matter.
Public Reference
CardX public article:
Technology Stack
| Category | Technologies |
|---|---|
| Cloud Platform | Customer-controlled AWS environment |
| Data Layer | Governed financial-data pipelines |
| Model Development | Offer and segmentation models with experiment-ready evaluation workflows |
| Optimization Workflow | Offer-combination evaluation and production-operational scoring flows |
| Governance | Controlled data handling inside the customer’s approved deployment perimeter |
