Overview

The Dataset task type lets your workflow store and retrieve key-value data that persists across runs. Use it to save state, accumulate results, or pass data between separate workflow executions — without needing an external database.

How it works

Dataset entries are scoped to a workflow. Each entry has a key and a value (which can be any JSON-compatible data). Entries persist across runs, so a value stored in one run is available in future runs of the same workflow.

Operations

Create – Store a new key-value pair. If the key already exists for the current workflow run context, it is updated.
Read by key – Retrieve the value for a specific key.
Read latest – Retrieve the most recently created entry.
Read latest N – Retrieve the N most recent entries.

Adding a dataset task

In the workflow builder, add a new task and select Dataset as the type.
Choose the operation (create, read by key, read latest, or read latest N).
Map the key and value from trigger inputs or previous task outputs using workflow references.
Use the task's output in downstream tasks as needed.

Use cases

Store a running total or counter across scheduled workflow runs.
Cache the result of an expensive API call for reuse in later runs.
Track the last processed record ID when syncing data from an external source.
Save intermediate results that another workflow can read via workflow chaining.

Tips

Use clear, descriptive keys (e.g. last_sync_timestamp, monthly_total) for readability.
Dataset storage is designed for lightweight state — not large datasets. For bulk data, use a database integration like PostgreSQL or MongoDB.