AI and iGaming Come Together [Case Study]
- Case Studies
Symphony Solutions helps Graphyte launch a new service, the only recommendation platform for the iGaming industry
Symphony Solutions is not new to the iGaming industry. Some of their largest clients are behemoths in this category with decades-long leadership. For those clients, Symphony Solutions provides fully-fledged sportsbook platforms for their clients’ growing number of punters. But for Graphyte, Symphony Solutions provides a different realm of collaboration.
Graphyte is a UK-based start-up SaaS company selling their services to large gambling and betting operators, which they (gaming clients) integrate with websites and mobile apps to offer personalized content to their users. Their main product, Graphyte Recommend, is the only recommendation platform for the iGaming industry.
It uses AI to provide ultra-personalized recommendations for bettors, much the same way Netflix provides entertainment program recommendations to their viewers. Graphyte receives a subscription fee from their clients, who in turn, increase user engagement and retention.
“Symphony Solutions helped us go from POC to a battle-hardened production solution in such a short time. Our current traction in the marketplace is testament to their great work.”Rob Davis, Co-Founder of Graphyte
Graphyte was a small start-up in October 2018 when they came to Symphony Solutions, having developed PoC, including public website, a back-office website, ML models, and a service to enable iGaming operators to recommend personalized content. Graphyte had been demonstrating their system to potential clients when they approached Symphony Solutions. They needed help in solving technical challenges with PoC, primarily scale and database performance related issues.
Symphony Solutions evaluated the situation and outlined immediate problems:
- Lack of development/demo/production environments
- Manual release process, manual changes without version control
- Lack of resources for system maintenance
- A bottleneck created by relational database in the center of the system
- Unnecessary latency
But these were just the Level 1 technical issues that needed to be solved.
The first level issues were addressed:
- Implementation of automatic infrastructure deployment by following laaC principle,
- Development of pipeline for Continuous Integration,
- Moved all application code under source control system
- Transferred system components to microservices
- Switched from relational database to DynamoDB database and streaming data processing
- Designed batch processing layer using Data Lake technologies, such as Athena, S3, Kenesis Firehose, AWS Glue, and AWS Batch
The bigger problems were ML/AI modeling scaling issues that accompany rapid growth in a short time. The teams worked quickly:
- A Machine Learning engineer was added to the team,
- PoC research was conducted to build a generic model for all clients in sports and gaming domains,
- Technical gaps in the production system were addressed to enable the Graphyte ML/AI engine to accommodate real-time inventory updates.
The Grahpyte and Symphony Solutions collaboration went from demo to production in a very short time, specifically:
- Created, improved and launched the service,
- Created barrierless integration for ease of client use,
- Improved code readability for maintainability,
- Provides personalized recommendations in 300ms end to end,
- Grew client brands from 1 to 9, which accommodate more than 50,000 active client punters per month,
- Projected compute cost reduction by 50%.
As this service and their products continue to penetrate, the business will be poised for extended growth. As more results are produced, this case study will be updated.