Collaborative web applications benefit from good responsiveness. This can be difficult to achieve with deployments on core data centers subject to high network latencies. Hybrid deployments using a mix of core and edge resources closer to end users are a promising alternative. Many challenges are associated with hybrid deployments of applications, starting from their decomposition into components able to be replicated dynamically onto edge resources to the management and consistency of these components’ state. We report on our experience with the hybrid deployment of ShareLatex, a legacy collaborative web application. We show how its design based on the use of microservices and resource-oriented APIs allow for an efficient modular decomposition. We detail how we adapted the application configuration for a hybrid deployment with no modification to its source code. Our experiments using a fleet of emulated users show that the use of a hybrid deployment for this legacy collaborative application can decrease user-perceived application latencies for common operations at the cost of increasing them for operations involving core/edge coordination traffic.