Glossary
The words SwiftTune uses for the parts of an AI application, defined once and linked from everywhere else.
- Flow
- A flow is the unit SwiftTune composes and deploys: a directed graph of blocks whose connections carry data from a trigger to a response.
- Block
- A block is one node of a flow: a named, configurable step that takes typed input on its input ports and produces typed output on its output ports.
- Port
- A port is a named, typed connection point on a block.
- Deployment
- A deployment is one published version of a flow: the Worker that Cloudflare runs, the URL it answers on, and the record of which flow version it came from.
- Edge
- The edge is the network of data centres a request reaches first, close to the person making it.
- Workers AI
- Workers AI is Cloudflare’s inference service: open models hosted on the same network the rest of a flow runs on, called through a binding instead of over the public internet.
- AI Gateway
- AI Gateway is Cloudflare’s proxy in front of model providers.
- Vectorize
- Vectorize is Cloudflare’s vector database: it stores embeddings and answers similarity queries over them.
- Retrieval-augmented generation (RAG)
- Retrieval-augmented generation is the pattern of answering a question by first retrieving relevant documents and then asking a model to answer using them.
- Evaluation
- An evaluation scores a flow against a fixed dataset of inputs and expected outcomes, so a change can be compared with the version before it.
- EU data residency
- EU data residency means the durable copies of your data are held in data centres inside the European Union.
- Bring your own account
- Bringing your own account means the flows you publish run in your Cloudflare account rather than in a pool the vendor owns.