> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryblend.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Build, test, publish, and reuse AI workflows.

Blend AI is a visual canvas for making AI workflows. You connect models, prompts, files, tools, choices, and User Result nodes, then run the workflow to see what happens.

The main use case is not writing integration code first. It is building a workflow you can test, compare, improve, and reuse.

There are two primary ways to use workflows:

<CardGroup cols={2}>
  <Card title="Experiment on the canvas" icon="diagram-project" href="/workflow/overview">
    Build workflows on the Gen Canvas, try different models and prompts, batch variations, compare outputs, and keep iterating.
  </Card>

  <Card title="Publish a runner" icon="rocket" href="/workflow/review-and-publish">
    Turn a workflow into a reusable page. User Input nodes become form fields, User Result nodes become outputs, and public workflows can charge a creator fee.
  </Card>
</CardGroup>

The API is optional. When a published workflow needs to run from your app, script, backend, or scheduled job, the same published workflow can also be called over HTTP.

## How workflows work

<Steps>
  <Step title="Build a workflow">
    Open the [workflow canvas](https://tryblend.ai/workflow), add nodes, connect handles, and configure the model, prompt, file, tool, or utility settings each step needs.
  </Step>

  <Step title="Run and compare">
    Run one node while designing, or run the whole workflow. Use batches and choice nodes to test multiple prompts, models, files, or outputs side by side.
  </Step>

  <Step title="Expose inputs and results">
    Add User Input nodes for values people should fill in later, and User Result nodes for outputs the workflow should return.
  </Step>

  <Step title="Publish when it is reusable">
    Review the workflow, choose visibility and pricing, add examples, then publish a runner page. If you need automation, use the generated API reference.
  </Step>
</Steps>

## Core concepts

<CardGroup cols={2}>
  <Card title="Canvas" icon="object-group" href="/workflow/overview">
    The workspace where you add nodes, wire handles, run experiments, and shape a reusable workflow.
  </Card>

  <Card title="Inputs and results" icon="right-left" href="/workflow/review-and-publish">
    User Inputs become runner form fields. User Results become returned output fields.
  </Card>

  <Card title="Lists & single values" icon="list-tree" href="/concepts/lists-and-values">
    How one value, many values, batches, selections, and LLM list outputs change how often downstream nodes run.
  </Card>

  <Card title="Output parts" icon="shapes" href="/parts">
    How completed runs return text, image, video, file, and JSON parts.
  </Card>
</CardGroup>

## Optional API

Every published workflow has an API surface. Use it when the workflow should run outside the Blend AI UI.

```bash theme={null}
curl -X POST https://tryblend.ai/api/v1/responses \
    -H "Authorization: Bearer bai_your_key_here" \
    -H "Content-Type: application/json" \
    -d '{
        "workflowId": "pub_abc123",
        "stream": true,
        "inputs": {
            "prompt": "A cat on the moon"
        }
    }'
```

<CardGroup cols={2}>
  <Card title="Build your first workflow" icon="diagram-project" href="/workflow/overview">
    The canvas, nodes, handles, data types, and runnable chains.
  </Card>

  <Card title="API Reference" icon="code" href="/api-reference/overview">
    Full endpoint reference with request/response schemas and a live playground.
  </Card>
</CardGroup>