FanDuel shrinks AWS migration time with AI

The challenge

Migrations to cloud architecture have become standard business practice across many sectors. This is due to the improved scalability they bring, allowing faster response to changing customer demand, along with lower hardware and maintenance costs compared with on-premises systems. This is proving particularly popular within gaming and gambling, as it means that during big events, for example, like the World Cup, betting platforms won't slow down just because millions of users log in at the same time. This is critical to providing the best possible customer experience.

Global betting company FanDuel has been working closely with decade-long technology partner Mindera on migration to the AWS Cloud. Mindera has found that it usually takes between 42 and 135 active days to migrate each service, which uses up valuable software engineer time. To accelerate the process, reducing the burden on engineers, and enabling the benefits to be realised more quickly, Mindera deployed artificial intelligence (AI) for the migration of the FanDuel’s Sportsbook betting platform to a new version of the AWS Cloud.

The solution

Mindera supported the migration by modernising infrastructure management (shifting from Chef to modern Ansible), moving deployments to pipelines, and improving availability and scalability through Auto Scaling Groups. AI-assisted extensions developed by FanDuel were incorporated into the workflow. Built on top of AI-powered integrated development environment Cursor, the AI extensions automate repetitive migration tasks, such as configuration translation, infrastructure setup and repository updates. Meanwhile, engineers retain responsibility for validation, approvals and production outcomes.

The solution combines off-the-shelf tooling with custom development. Cursor provides the development environment, while Spec-Kit AI extensions, which structure AI-assisted software development through Spec-Driven Development (SDD), are tailored to FanDuel’s infrastructure patterns and migration requirements. This drives clarity on expected AI behaviour before implementation, reducing issues like scope creep, tech debt and hallucinations in AI-assisted workflows. It also makes the process reusable for similar migration workloads, with adaptation to each domain.

The results

This approach improves throughput without increasing team size. As a result, migrations were completed faster and in parallel, with more than eight engineers onboarded and productive within the model. Furthermore, standardised patterns allow engineers to work independently while producing consistent outputs.

This has enabled AWS migrations to be completed eight times faster, shrinking the process to between four and 25 days per service, with the main gains coming from automating pattern-based work. This includes:

• Migrating from open-source IT automation and configuration management tool Chef to Ansible.
• Unifying critical information by generating Datadog catalogues.
• Setting up load balancers and applying consistent change across repositories.

The extensions also include verification steps that identify issues earlier in the process, reducing rework.

Consistency is another key outcome. Generated artifacts such as Ansible roles, observability configurations, and infrastructure components follow the same structure across services. This reduces variability between engineers and limits configuration drift.

From an engineering perspective, effort has shifted away from repetitive manual tasks toward validation and exception handling. Engineers spend less time rewriting configuration or setting up infrastructure, and more time ensuring correctness and handling non-standard cases. Onboarding is also faster, with new engineers able to complete their first migration in days rather than months.

Important limitations of AI-assisted migration were also discovered. Early migrations exposed issues with context handling, recurring errors and onboarding friction. AI assistance was found less effective when requirements are unclear or when migrations fall outside established patterns. These challenges are being addressed through improved guardrails and better specification of migration patterns.

Overall, AI-assisted migrations reduce time to migrate services, improve consistency, and increase delivery speed, while maintaining engineering control over risk and production quality.

Get in touch to discuss accelerating your cloud journeys…