Blocks are the canvas primitives, the atomic units of Spine. Think of them as Lego bricks for AI workflows: each one does something specific, but they can be snapped together and composed in many different ways. You can connect any block to any other block, and that connection guarantees the passing of context regardless of block type.Blocks fall into five categories:AI Output Blocks: Core reasoning and generation blocks: Chat, Prompt, Deep Research, MemoArtifact Builder Blocks: Produce downloadable, shareable, production-quality deliverables. More agentic than other blocks, running models in loops behind the scenes: Slides, Document, Excel, App (including Prototype and Landing Page variations), Image Generation, FramesIntermediary Blocks: Produce structured content used as input for downstream blocks or as working artifacts: Table, ListAgentic / Execution Blocks: BrowserUse (spins up real browser sessions to interact with the web)Data and Source Ingestion Blocks: Bring external content into the canvas: Web Page, YouTube, Files, Folder, Folder Synthesis, InputsUser and Agent Annotation Blocks: Notes (free-form text for human or agent annotations)You can draw arrows from blocks to pass context from one block to another. Not all blocks share the same buttons. Below are what all buttons do on blocks.
Commonly found on image & prompt blocks — allows the user to create variants of the block using the same or different model. Learn more about bulk variant generation.
