How Future Anthem Scales AI Personalisation on AWS

The migration of Future Anthem’s real-time personalisation architecture to Amazon Web Services (AWS) wasn’t just about shifting workloads. 

This is a fundamental redesign, providing us with a strategic opportunity to re-engineer Amplifier AI from the inside out — to build better, move faster, unlock broader access and reduce costs.

Importantly, we further modularised the platform to enhance accessibility via the AWS Marketplace. This transformation enables faster development and deployment cycles, allowing customers to access innovation sooner. While Amplifier AI was already operating in the cloud, this re-architecture gave us the opportunity to rethink everything — from customer onboarding to machine learning deployment. All of this was achieved while maintaining the scale to process over 10 billion bets per month for operators worldwide.

How did we do it? Read on to find out.

1. Setting Our North Star: Defining the Holy Grail

We started by defining what an optimised, future-ready scenario should look like. For us, it centred around one-click deployments.

To get there, we focused on: 

• Reviewing the process for account, resource and security creation 

• Mapping every data flow from ingestion through to deployment 
• Isolating shared components into reusable services, with API boundaries for future flexibility 
• Further abstracting our internal ‘Machine Learning as Software’ patterns 
  

2. Building the Backbone: Unifying Our Frameworks

Amplifier AI has been fundamentally redesigned to enable seamless communication across its product suite. This ensures that player decisions are made holistically, across the whole player journey. With standardised APIs and service boundaries, operators can adopt the specific capabilities they need, when they need them.

This modular approach ensures flexible scaling and customisation as operational needs evolve. We all know the value of clean, efficient architecture. For this project, we followed three clear principles:

• Only migrate standard patterns — any gaps were closed during the migration, ensuring a clean start 
• Invest in automation — standardised automation now enables faster scaling across the customer journey 
• Invest in abstraction – every piece of production code was reviewed against an objective of one-click deployment

3. Thinking Ahead: Designing for the Future  

With Amplifier AI processing billions of transactions monthly, we had clear visibility into where optimisation would have the most impact. We focused on: 

• Migrating to AWS-native data services for speed, scalability, resilience and simplified support 

• Embracing data lineage and AI system traceability, with Unity Catalog ensuring full transparency from ingestion to live models, in line with emerging legislation

• Enhancing real-time monitoring of latency, throughput and data quality 
• Prioritising cost optimisation through efficient service design and fine-grained resource tagging for analysis

4. Switching Without a Hitch: Managing Client Cutover 

The live migration was one of the most critical phases — and getting it right meant eliminating risk without compromising player experiences. 

Here’s how we did it: 

• Provisioned parallel AWS environments to mirror production workloads 

• Synced live data streams to keep systems up to date throughout 

• Rolled out cutovers, module-by-module, reducing risk and allowing incremental validation 
• Verified performance and accuracy immediately after switching over 
• Worked very closely with internal and client teams communicating at every step

This ensured a smooth, seamless transition, with zero disruption for operators and players alike.

We also held a series of workshops with AWS's Betting and Gaming specialists before and after the migration - ensuring continuous collaboration with their team throughout the process.

5. Gaining Control: Embedding Observability and Resilience 

The new architecture was purpose-built for transparency, control, and reliability. 

• Centralised dashboards give real-time visibility into performance and usage metrics 
• Automated alerts track anomalies, model drift, and any signs of degradation 
• Rollbacks are built into every module to enable confident, rapid deployment

6. Launching at Pace: Available via the AWS Marketplace  

By improving how Amplifier AI is packaged and deployed, we’ve made it easier for new customers to onboard — listing its capabilities directly on the AWS Marketplace. 

This step streamlines discovery, purchasing and integration. It also opens new co-selling opportunities worldwide through our inclusion in the AWS ISV Accelerate Programme.

Future Anthem is now also listed within AWS Betting and Gaming Partner network and solutions library.

Explore Future Anthem on the AWS Marketplace

The Outcome: Best-in-Class, Reimagined  

This wasn’t just a migration – it was a strategic re-architecture that has delivered stunning results: 

• Cutting cloud costs by 50% 
• Cutting response times of already performant APIs even further, now under 100ms
• Enabling automated (with human in the loop) onboarding of new clients
• Preserving the ability to process hundreds of billions of spins 
• Making Amplifier AI easier to access and adopt via AWS Marketplace
• Strengthening security, observability and operational control

As a result, we’ve created a platform that operators can deploy faster, tailor to their needs and scale without limits. 

We're continuing to evolve the platform, with further optimisation underway to enhance performance and user experience.

See it in Action 

Book a demo to explore how Amplifier AI delivers AI-driven personalisation at scale.