Dray VS GraphPilot

GraphPilot's answer:

GraphPilot is built by Kay Schecker, a solo founder from Düsseldorf, Germany. Caching GraphQL at the edge is easy to start and hard to get right: most setups end up choosing between short TTLs and purging far too much. GraphPilot was built to use the schema itself as the source of truth for what a response contains, so invalidation can be precise. The slogan says it: "We fly the complexity so you don't have to."

GraphPilot's answer:

Choose GraphPilot if you want caching that is driven by your schema and invalidation you can verify. Setup is a config file plus your schema, and you are live in minutes. It works with Apollo Server, GraphQL Yoga, PostGraphile and federation routers, and it also caches plain HTTP APIs. It runs on Fastly's global edge network, is built and operated in Germany, and has a free plan (Glider) so you can test it on real traffic before you pay.

GraphPilot's answer:

GraphPilot reads your GraphQL schema and turns it into cache keys. It adds the key fields each query needs, removes them from the response again, and tags every cached response with the objects it contains. When a mutation returns a changed object, GraphPilot purges the responses that contain it, without hand-written cache tags or short TTLs as a safety net. You can see each purge, with its reason, in the Cockpit.

GraphPilot's answer:

Engineering teams running GraphQL in production with read-heavy traffic: e-commerce, marketplaces, content platforms, SaaS dashboards and mobile app backends. Typically a platform or backend team that wants less origin load and lower latency without building and maintaining its own caching and invalidation layer.

GraphPilot's answer:

The edge runtime is written in Rust and compiled to WebAssembly on Fastly Compute. The platform and API run on TypeScript and PostgreSQL, the Cockpit is built with React, and the CLI (gpilot) is a single Rust binary for macOS, Linux and Windows.