workflow/docs/api-reference/workflow-serde/workflow-deserialize.mdx

Workflow SDK - Build durable, resilient, and observable workflows

---
title: WORKFLOW_DESERIALIZE
---

A symbol used to define custom deserialization for user-defined class instances. The static method should accept serialized data and return a new class instance.

## Usage


```typescript lineNumbers
import { WORKFLOW_SERIALIZE, WORKFLOW_DESERIALIZE } from "@workflow/serde";

class Point {
  constructor(public x: number, public y: number) {}

  static [WORKFLOW_SERIALIZE](instance: Point) {
    return { x: instance.x, y: instance.y };
  }

  static [WORKFLOW_DESERIALIZE](data: { x: number; y: number }) {
    return new Point(data.x, data.y);
  }
}
```

## API Signature

{/* @skip-typecheck: type-only signature snippet, not compilable code */}

```typescript
static [WORKFLOW_DESERIALIZE](data: SerializableData): T
```

### Parameters

<TSDoc
definition={`
interface Parameters {
  /**
   * The serialized data to reconstruct into a class instance.
   * This is the same data that was returned by WORKFLOW_SERIALIZE.
   */
  data: SerializableData;
}
export default Parameters;`}
/>

### Returns

The method should return a new instance of the class, reconstructed from the serialized data.

## Requirements

<Callout type="warn">
The method must be implemented as a **static** method on the class. Instance methods are not supported.
</Callout>

- Both `WORKFLOW_SERIALIZE` and `WORKFLOW_DESERIALIZE` must be implemented together
- The method receives the exact data that was returned by `WORKFLOW_SERIALIZE`
- If `WORKFLOW_SERIALIZE` returns complex types (like `Map` or `Date`), they will be properly deserialized before being passed to this method

<Callout type="warn">
This method runs inside the workflow context and is subject to the same constraints as `"use workflow"` functions:
- No Node.js-specific APIs (like `fs`, `path`, `crypto`, etc.)
- No non-deterministic operations (like `Math.random()` or `Date.now()`)
- No external network calls

Keep this method simple and focused on reconstructing the instance from the provided data.
</Callout>