Next Steps

You now understand the basic WOML development cycle. The best next page depends on the system you want to build, so use this guide as a learning roadmap rather than reading every reference page in order.

Build an application endpoint

Choose Webhooks when an application or external service must submit work over HTTP. The tutorial covers payload schemas, authentication, generated curl commands, asynchronous acceptance, and the difference between a rejected request and a failed run.

Recommended path:

Build scheduled automation

Use Schedules and Intervals for reports, maintenance, synchronization, and recurring checks. Learn missed-run behavior and timezones before deploying a calendar-sensitive process.

Build an AI or communication workflow

Start with Communication Triggers, then configure the provider you need. A provider message becomes normalized context.payload, so the workflow can call an LLM through HTTP or a local module, use durable state, request human approval, and send a response.

Recommended path:

Model decisions, loops, and concurrency

Read Choose the Right Primitive before adding control flow. It explains when to use a step, choice, switch, parallel group, durable loop, fork, or approval.

The usual progression is:

Sequential steps for dependent work.
<parallel> for independent single operations.
<choose> or <switch> for one selected route.
<for-each> for durable work over a runtime array.
<fork> for independent multi-step lanes.

Connect systems and preserve data

Services Overview introduces supervised HTTP, databases, object storage, cache, durable state, events, workflow calls, and messaging.

Choose storage by meaning: database for application records, storage for larger objects, cache for disposable optimization, and state for small workflow-owned memory that must survive future runs.

Reuse and compose

Move ordinary transformations and API wrappers into JavaScript and TypeScript Modules. Create Reusable WOML Definitions when the reusable unit itself needs durable step behavior, props, results, or lifecycle observation.

Use Workflow Communication when processes should remain independent. A workflow can wait for a child's result, start background work, or publish an event to many subscribers.

Prepare for production

Before exposing a workflow to real traffic, work through:

Production readiness is not only “the workflow ran once.” It means failures are understandable, effects are safe to retry, secrets remain outside source, state is backed up, and an operator can inspect or stop the runtime.