About the Playground

the playground is a high fidelity, live e commerce platform designed to mirror the challenges of modern production systems it’s where you can build intuition for resolve ai’s reasoning and data correlation capabilities without high stakes pressure why use the playground? production environments are complex and high stakes whether it’s 3 am during an on call rotation or a mid day deployment, you don’t want to be learning a new tool in the heat of an incident the playground allows you to practice natural language querying learn what questions resolve ai can answer data correlation see how it links logs, metrics, and code root cause analysis observe how the ai reasons through distributed system failures system architecture behind the storefront is a sophisticated microservices architecture customers browse, add to carts, and checkout, while 19 backend services coordinate the experience playground architecture diagram key service groups the shopping flow product catalog , recommendation , cart , and checkout business logic payment , fraud detection , accounting , and shipping foundational services currency , email , quote , and ad infrastructure kafka (via aws msk), valkey cart (redis), and aws rds (postgresql) all services run on aws eks (us east 2), with dozens of pods instrumented with opentelemetry the observability stack resolve ai has deep visibility into the playground's telemetry when you ask a question, it queries these backends simultaneously to build a complete picture telemetry type backend what resolve ai sees logs loki errors, stack traces, and debug events metrics prometheus cpu/ram usage, request rates, and p99 latency traces tempo the full path of a single request across services infrastructure k8s api pod health, restart counts, and deployment state code github commit history and specific code changes life of a request understanding how traffic flows helps you investigate more effectively here is the typical path visit request hits the frontend browse frontend → product catalog for data cart items are stored in cart (backed by valkey cart ) checkout checkout coordinates with shipping , payment , and fraud detection fulfillment email confirms the order while accounting logs the transaction cascading failures in the playground, a slow payment service might cause a backup in checkout , just like in the real world resolve ai is designed to find these causal links what to expect continuous traffic a load generator ensures there is always data to analyze genuine issues failures aren't scripted; they're the result of real service interactions and occasional simulated chaos high availability most services run with multiple replicas to ensure the system stays up even during localized failures