Skip to main content

Your First Workflow

This walkthrough builds a complete workflow from scratch: it receives a piece of text as input, summarises it using an AI prompt, and sends the result to a Slack channel.

By the end you'll have a working workflow with a real-world structure that applies to dozens of similar use cases.

Before you start​

You'll need:

  • A Slack connector connected on the Connectors page (see OAuth Connectors)
  • A Slack channel to send messages to

Step 1: Create the workflow​

Navigate to Workflows and click New Workflow. Give it a name - "Text Summariser to Slack".

The canvas opens with a Start node already placed.

Step 2: Define your input​

Click the Start node. Add one input variable:

  • Name: text
  • Type: string

This is the text the workflow will receive when it runs.

Step 3: Add a Prompt node​

Click the + on the Start node's output edge and select Prompt.

In the Prompt node configuration:

  • Select a prompt from your library, or click Create new prompt to write inline instructions
  • Map the text input variable to the prompt's variable

If creating inline, use instructions like:

Summarise the following text in 2-3 sentences:

{{text}}

Step 4: Add a Slack node​

Click + on the Prompt node's output and select Slack from the integrations list.

Configure the Slack node:

  • Operation: Post message
  • Channel: your target channel name or ID
  • Message: select the output variable from the Prompt node (the summary)

Step 5: Test it​

Click Run Workflow. Enter a sample text in the input field and run.

Check the execution results in the run panel - you'll see each node's input, output, and status. If the Slack message arrived in your channel, the workflow is working.

Step 6: Deploy it​

Go to Entry Points and create a Chat entry point to make it runnable from the chat launcher, or an API entry point to call it programmatically.

See Distributing and Scheduling Workflows for all entry point options.

From here you can extend this workflow - add a Condition node to skip Slack if the text is too short, use a Collection Loop to summarise multiple pieces of text in one run, or swap Slack for a different integration. The structure stays the same; the nodes change.