Common Patterns

Common distributed patterns are simple to implement in workflows and require learning no new syntax. You can just use familiar async/await patterns.

Sequential Execution

The simplest way to orchestrate steps is to execute them one after another, where each step can be dependent on the previous step.

export async function dataPipelineWorkflow(data: unknown) {
  "use workflow";

  const validated = await validateData(data);
  const processed = await processData(validated);
  const stored = await storeData(processed);

  return stored;
}

Parallel Execution

When you need to execute multiple steps in parallel, you can use Promise.all to run them all at the same time.

export async function fetchUserData(userId: string) {
  "use workflow";

  const [user, orders, preferences] = await Promise.all([
    fetchUser(userId),
    fetchOrders(userId),
    fetchPreferences(userId)
  ]);

  return { user, orders, preferences };
}

This not only applies to steps - since sleep() and webhook are also just promises, we can await those in parallel too. We can also use Promise.race instead of Promise.all to stop executing promises after the first one completes.

import { sleep, createWebhook } from "workflow";

export async function runExternalTask(userId: string) {
  "use workflow";

  const webhook = createWebhook();
  await executeExternalTask(webhook.url); // Send the webhook somewhere

  // Wait for the external webhook to be hit, with a timeout of 1 day,
  // whichever comes first
  await Promise.race([
    webhook,
    sleep("1 day"),
  ]);

  console.log("Done")
}

A Full Example

Here’s a simplified example taken from the birthday card generator demo, to illustrate how sequential and parallel execution can be combined.

import { createWebhook, sleep, type Webhook } from "workflow"

async function birthdayWorkflow(
    prompt: string,
    email: string,
    friends: string[],
    birthday: Date
) {
    "use workflow";

    // Generate a birthday card with sequential steps
    const text = await makeCardText(prompt)
    const image = await makeCardImage(text)

    // Create webhooks for each friend who's invited to the birthday party
    const webhooks = friends.map(_ => createWebhook())

    // Send out all the RSVP invites in parallel steps
    await Promise.all(
        friends.map(
            (friend, i) => sendRSVPEmail(friend, webhooks[i])
        )
    )

    // Collect RSVPs as they are made without blocking the workflow
    let rsvps = []
    webhooks.map(
        webhook => webhook
            .then(req => req.json())
            .then(( { rsvp } ) => rsvps.push(rsvp))
    )

    // Wait until the birthday
    await sleep(birthday)

    // Send birthday card with as many rsvps were collected
    await sendBirthdayCard(text, image, rsvps, email)

    return { text, image, status: "Sent" }
}

Timeout Pattern

A common requirement is adding timeouts to operations that might take too long. Use Promise.race with sleep() to implement this pattern.

import { sleep } from "workflow";

export async function processWithTimeout(data: string) {
  "use workflow";

  const result = await Promise.race([
    processData(data),
    sleep("30s").then(() => "timeout" as const),
  ]);

  if (result === "timeout") {
    // In workflows, any thrown error exits the workflow (FatalError is for steps)
    throw new Error("Processing timed out after 30 seconds");
  }

  return result;
}

This pattern works with any promise-returning operation including steps, hooks, and webhooks. For example, you can add a timeout to a webhook that waits for external input:

import { sleep, createWebhook } from "workflow";

export async function waitForApproval(requestId: string) {
  "use workflow";

  const webhook = createWebhook<{ approved: boolean }>();
  await sendApprovalRequest(requestId, webhook.url);

  const result = await Promise.race([
    webhook.then((req) => req.json()),
    sleep("7 days").then(() => ({ timedOut: true }) as const),
  ]);

  if ("timedOut" in result) {
    throw new Error("Approval request expired after 7 days");
  }

  return result.approved;
}

Workflow Composition

Workflows can call other workflows, enabling you to break complex processes into reusable building blocks. There are two approaches depending on your needs.

Direct Await (Flattening)

Call a child workflow directly using await. This “flattens” the child workflow into the parent - the child’s steps execute inline within the parent workflow’s context.

// Child workflow
export async function sendNotifications(userId: string) {
  "use workflow";

  await sendEmail(userId);
  await sendPushNotification(userId);
  return { notified: true };
}

// Parent workflow calls child directly
export async function onboardUser(userId: string) {
  "use workflow";

  await createAccount(userId);
  await sendNotifications(userId);
  await setupPreferences(userId);

  return { userId, status: "onboarded" };
}

With direct await, the parent workflow waits for the child to complete before continuing. The child’s steps appear in the parent’s event log as if they were called directly from the parent.

Background Execution via Step

To run a child workflow independently without blocking the parent, use a step that calls start(). This launches the child workflow in the background.

import { start } from "workflow/api";

// Step that starts a workflow in the background
async function triggerReportGeneration(reportId: string) {
  "use step";

  const run = await start(generateReport, [reportId]);
  return run.runId;
}

// Parent workflow
export async function processOrder(orderId: string) {
  "use workflow";

  const order = await fulfillOrder(orderId);

  // Fire off report generation without waiting
  const reportRunId = await triggerReportGeneration(orderId);

  // Continue immediately - report generates in background
  await sendConfirmation(orderId);

  return { orderId, reportRunId };
}

With background execution, the parent workflow continues immediately after starting the child. The child workflow runs independently with its own event log and can be monitored separately using the returned runId. Choose direct await when:

The parent needs the child’s result before continuing
You want a single, unified event log

Choose background execution when:

The parent doesn’t need to wait for the result
You want separate workflow runs for observability

Getting Started

Installation

Core Concepts

Building AI Agents

How It Works

Deployment

Observability

Testing

Error Reference

Sequential Execution

Parallel Execution

A Full Example

Timeout Pattern

Workflow Composition

Direct Await (Flattening)

Background Execution via Step

​Sequential Execution

​Parallel Execution

​A Full Example

​Timeout Pattern

​Workflow Composition

​Direct Await (Flattening)

​Background Execution via Step

Sequential Execution

Parallel Execution

A Full Example

Timeout Pattern

Workflow Composition

Direct Await (Flattening)

Background Execution via Step