Getting Started

Build a Call-to-Slack Summary Workflow in 5 Minutes

Turn sales call transcripts into Slack summaries with AI

After every sales call, you spend 15 minutes writing notes, updating the CRM, and pinging your team about next steps.

What if it just… happened?

Your call recorder probably has a built-in summarizer—but it’s a black box. You can’t customize the prompts, pick a better model, or do anything with the output. When you build your own, you control exactly what gets extracted, how it’s formatted, and where it goes next.

In this guide, you’ll build a simple workflow that takes a call transcript and posts a summary to Slack. Along the way, you’ll learn the fundamentals—inputs, blocks, dependencies, and outputs—that power every Scout automation. Once you’ve got the basics, you can extend this workflow or build something entirely new.

1

Create your workflow

  1. Sign up free or sign in
  2. Click Studio in the sidebar, then click Workflows
  3. Click + New to create a new workflow
  4. Name it “Call Summary Bot” and click Create

You’ll see an empty workflow canvas with a “Set a trigger” button.

2

Set up the Input

  1. Click Set a trigger
  2. Click Input from the list
  3. In the settings panel on the right, change the ID from message to transcript
  4. Click anywhere on the canvas to close the settings

This creates an input field where you’ll paste your call transcripts.

The ID you set here (transcript) is how you’ll reference this input in other blocks: {{ inputs.transcript }}

3

Add the LLM block

This block will analyze the transcript and extract the important stuff.

  1. Click the + button at the bottom of the canvas
  2. Search for LLM and drag it onto the canvas
  3. Connect the blocks: Draw a line from the bottom of the Input block to the top of the LLM block

This line declares a dependency—it tells Scout that the LLM block should run after the Input block, and can access its data. Blocks without connections run in parallel.

  1. Click the LLM block to open its settings
  2. Configure it:
    • Model: gpt-4o
    • System message: 
      You analyze sales call transcripts. Extract:
      📋 **Summary** (3 sentences max)
      💬 **Key Discussion Points** (bullets)
      ✅ **Action Items** (with owners if mentioned)
      ➡️ **Next Steps**
      Be concise. Use Slack-friendly markdown formatting.
    • User message: 
      1{{ inputs.transcript }}
4

Test it out

Before adding more blocks, let’s see it work! You can test anytime using the debug console.

  1. Click the ▶ Play button in the top left to open the Console

  2. Paste this sample transcript:

    Sample Transcript
    Rep: Thanks for taking the time today. So you mentioned you're currently
    using spreadsheets to track your sales pipeline?
    Prospect: Yeah, it's a mess honestly. We have like 50 reps and zero
    visibility into what's happening.
    Rep: Totally get it. What we could do is set up a pilot with 5 reps,
    get you connected to your CRM, and you'd see results within 2 weeks.
    Prospect: That sounds reasonable. Can you send over pricing?
    Rep: Absolutely. I'll send that over today and let's reconvene Thursday
    to discuss. I'll send a calendar invite.
    Prospect: Perfect, thanks.

  3. Click Run

  4. Watch the LLM extract a clean summary with action items ✨

The debug console lets you test your workflow at any point—no need to publish first.

5

Add the Slack block

Now let’s send that summary to your team automatically.

  1. Click the + button at the bottom of the canvas
  2. Search for Post Slack Message and drag it onto the canvas
  3. Connect the blocks: Draw a line from the bottom of the LLM block to the top of the Slack block
  4. Click the Slack block to configure it:

    • Slack Workspace: Select your workspace, or click + Add Workspace to connect Slack

    • Channel: Select your channel (e.g., #sales-calls)

    • Message: {{ llm.output }}

      Replace llm with your actual block ID—find it by hovering over the LLM block and clicking the copy icon.

First time with Slack? After connecting your workspace, add the Scout bot to your channel by typing @Scout and clicking Add to Channel.

6

Publish and run

Your changes are currently in draft mode—only you can see them. Publishing creates a new version that can be used by triggers and integrations.

  1. Click Publish in the top right
  2. Open the Console again and run it with a transcript
  3. Watch the summary appear in Slack!

What You Just Built

You created a workflow that:

  • Takes input → the transcript
  • Processes it with AI → LLM extracts summary + action items
  • Sends it somewhere useful → Slack channel

This is the core pattern for most Scout automations: Input → Process → Output.

This simple workflow is just the starting point. From here, you can add more blocks, chain multiple AI calls, branch with conditionals, or connect to other services. The same fundamentals you just learned apply to everything you’ll build in Scout.

What’s Next?

You’ve got the basics. Now make it powerful:

Build an Agent

Create an AI assistant that can use tools and make decisions

Update HubSpot

Add deal notes to your CRM automatically

Build a RAG App

Create an AI that answers questions using your data

Explore All Blocks

See everything you can build with

Want to skip the building? Check out our Workflow Templates to start from pre-built automations.