AI Workflows
Chain multi-step AI pipelines across providers. LLM calls, image generation, data transforms, and HTTP requests.
$0.001 / step
- Multi-Provider
- Step Types
- Variables
// Define a content pipeline workflowconst workflow = await stack0.workflows.create({name: 'content-pipeline',steps: [{id: 'research',type: 'llm',provider: 'anthropic',model: 'claude-sonnet-4-20250514',prompt: 'Research {{variables.topic}} and list key points',},{id: 'draft',type: 'llm',provider: 'openai',model: 'gpt-4o',prompt: 'Write a blog post using: {{research.output}}',},],})const result = await stack0.workflows.runAndWait({workflowId: workflow.id,variables: { topic: 'AI agent infrastructure' },})
Built for production
LLM Steps
Call any LLM with prompt templates and output parsing.
{type: 'llm', provider: 'anthropic', model: 'claude-sonnet-4-20250514',}
Call any LLM with prompt templates and output parsing.
Image Generation
Generate images with Flux, SDXL, or Stable Diffusion.
{type: 'image', provider: 'replicate', model: 'flux-schnell',}
Generate images with Flux, SDXL, or Stable Diffusion.
HTTP Requests
Call external APIs and incorporate responses into workflow.
{type: 'http', method: 'POST', url: '{{variables.apiUrl}}',}
Call external APIs and incorporate responses into workflow.
Data Transforms
Parse, filter, and transform data between steps.
{type: 'transform', operation: 'map', path: 'items[*].name',}
Parse, filter, and transform data between steps.
Conditionals
Branch workflow based on conditions and comparisons.
{type: 'condition', if: '{{research.length}} > 5',}
Branch workflow based on conditions and comparisons.
Webhooks
Receive results via webhook when workflow completes.
{webhookUrl: 'https://yourapp.com/webhook',}
Receive results via webhook when workflow completes.
Multi-Provider
Use Anthropic, OpenAI, Gemini, Replicate, and more in the same workflow.
Step Types
LLM calls, image generation, HTTP requests, data transforms, and conditionals.
Variables
Pass data between steps with interpolation syntax like {{step1.output}}.
Retries
Automatic retry with exponential backoff on transient failures.
Parallel Steps
Run independent steps in parallel for faster execution.
Conditionals
Branch workflow logic based on previous step outputs.
Built for real applications
Content Pipelines
Generate outlines, write drafts, edit, and publish—all in one workflow.
Simple, usage-based
$20/mo includes $10 of usage. Metered beyond that, at the same rate for everyone.
Steps
$0.001/ step
- LLM calls (+ provider cost)
- Image generation (+ provider cost)
- HTTP requests
- Data transforms
- Conditionals
- Automatic retries
- Webhook notifications
Frequently asked questions
Anthropic (Claude), OpenAI (GPT-4), Google (Gemini), Replicate (Flux, SDXL), Together (Llama, Mixtral), and more. Use different providers for different steps.
Each step costs $0.001 to execute, plus underlying provider costs (OpenAI tokens, Replicate compute, etc.). Simple, transparent pricing.
Yes. Use {{stepId.output}} interpolation to reference outputs from previous steps. Variables flow through the entire workflow.
Automatic retry with exponential backoff for transient failures. Configure max retries and fallback behavior per step.
Yes. Steps without dependencies run in parallel automatically. Explicitly declare dependencies to control execution order.
Yes. Start a workflow, receive a job ID, and poll for completion or receive results via webhook.
Ready to build?
Get started in minutes.