# Blend AI Docs

- [Introduction](https://docs.tryblend.ai/introduction.md): Build, test, publish, and reuse AI workflows.
- [Workflow Overview](https://docs.tryblend.ai/workflow/overview.md): How to build a workflow: the canvas, nodes, connections, and data types.
- [Review and Publish](https://docs.tryblend.ai/workflow/review-and-publish.md): Turn a workflow into a reusable runner page, share it, and optionally expose it through the API.
- [Data types & compatibility](https://docs.tryblend.ai/concepts/data-types.md): What each handle's data type means, and the rules for which connections are allowed.
- [Handles, edges & connections](https://docs.tryblend.ai/concepts/handles-and-edges.md): The dots you wire together, the connections they form, and the rules that govern what's allowed.
- [Lists & single values](https://docs.tryblend.ai/concepts/lists-and-values.md): Why a single value runs the next node once and a list runs it once per item — plus items vs. parts, structured-list fan-out, failures, and the API shape.
- [Output Parts](https://docs.tryblend.ai/parts.md): Every workflow result is an array of output items, and every item contains typed parts.
- [Running a single node](https://docs.tryblend.ai/concepts/single-node-run.md): Run one node on its own to test it, with its results kept right on the node.
- [Running the full workflow](https://docs.tryblend.ai/concepts/full-run.md): Run the saved graph from the editor as a durable, recorded workflow run.
- [The Runner](https://docs.tryblend.ai/concepts/runner.md): The published, form-driven front door for rerunning a workflow on demand.
- [Results & history](https://docs.tryblend.ai/concepts/results-and-history.md): Where a result lives depends on how you ran it — on the node, or as a run in history.
- [Editor vs Runner](https://docs.tryblend.ai/concepts/editor-vs-runner.md): The editor and the Runner are two surfaces over the same workflow — one for building, one for rerunning.
- [Agents, Sub-agents & Tools](https://docs.tryblend.ai/workflow/agents-and-tools.md): The difference between an Agent, a Sub-agent, and a Tool — and how to wire them together.
- [User Input](https://docs.tryblend.ai/workflow/nodes/input.md): Exposes a value as a form field people fill in when running your published workflow, and as an input parameter in the API.
- [Batch](https://docs.tryblend.ai/workflow/nodes/batch.md): Fan a workflow out over prompts, models, parameters, or files—with explicit Set and Explode behavior for file groups.
- [Collection](https://docs.tryblend.ai/workflow/nodes/collection-read.md): Load a saved collection as one reusable file bundle, selected on the node or supplied at run time.
- [LLM](https://docs.tryblend.ai/workflow/nodes/llm.md): Runs a language model on a prompt. Works as a standard step (Agent) or as an on-demand helper other agents can call (Sub-agent).
- [Image Gen](https://docs.tryblend.ai/workflow/nodes/image-gen.md): Generate or edit images from a prompt and optional reference images, using the image model you pick.
- [Video Gen](https://docs.tryblend.ai/workflow/nodes/video-gen.md): Generates a video from a prompt and optional reference media using the model you select.
- [Tool](https://docs.tryblend.ai/workflow/nodes/tool.md): Bundles built-in tools for an agent to call, or runs a single tool on its own as a step. Two modes, one node.
- [Utility](https://docs.tryblend.ai/workflow/nodes/utility.md): Run a deterministic helper operation — extract audio from a video or compose text from a template.
- [Choose Output](https://docs.tryblend.ai/workflow/nodes/output-comparison.md): Pause the workflow and pick one or more upstream outputs to continue with.
- [User Result](https://docs.tryblend.ai/workflow/nodes/result.md): Marks a workflow output that's returned to API callers as a response field.
- [Annotation](https://docs.tryblend.ai/workflow/nodes/annotation.md): Add sticky notes, image boards, frames, and arrows to explain and organize a workflow without changing its execution.
- [Overview](https://docs.tryblend.ai/api-reference/overview.md): Quick tour of the Blend AI Workflow API.
- [Authentication](https://docs.tryblend.ai/api-reference/authentication.md): How to authenticate requests to the Blend AI Workflow API.
- [Streaming](https://docs.tryblend.ai/api-reference/streaming.md): Server-Sent Events format and how to consume them.
- [Errors](https://docs.tryblend.ai/api-reference/errors.md): Error response shape, status codes, and sub-codes.
- [Start a response](https://docs.tryblend.ai/endpoint/start-a-response.md): Starts a published workflow run.
- [Get a run](https://docs.tryblend.ai/endpoint/get-a-run.md): Returns current state of a run. Poll every 1-3 seconds until `status` is `completed`, `failed`, or `cancelled`.
- [Stream a run](https://docs.tryblend.ai/endpoint/stream-a-run.md): Opens an SSE stream for an existing run.

## OpenAPI Specs

- [openapi](/openapi.json)