Getting Started with Otto AI Agents: A Step-by-Step Workflow Example

Imagine your workflow automation not as a rigid set of rules, but as an intelligent assistant—one that thinks, learns, and makes decisions like a human.

Otto AI Agents aren’t just an upgrade—they’re the future of business automation. With human-like reasoning and contextual understanding, they bring unprecedented intelligence and adaptability to your workflows.

In this guide, you’ll walk through a powerful use case showcasing how to configure and use an Otto AI Agent from start to finish.

Visual Walkthrough

This setup video covers:

• Configuring the trigger
• Setting up the AI Agent
• Mapping dynamic fields
• Adding source actions
• Sending outputs (e.g., to Slack or Trello)
• Viewing results on the History page

Example Use Case

Goal:

When a user sends a message in a Slack channel—whether it’s reporting a bug or suggesting a new feature—the Otto AI Agent will:

1. Analyze the message content.
2. Create a well-formatted Trello task based on its type (Bug, Task, or User Story).
3. Reply in the same Slack thread with a confirmation message and the Trello task link.

Step 1: Add the Trigger

Trigger: Slack – New Channel Message
Why: This starts the workflow when a new message is posted in a Slack channel.

Important Note:
Set “Trigger for bot messages?” to No.
Since the AI later replies in the same thread using the Slack: Send Message (as Bot) action, enabling bot-triggered messages would cause an infinite loop.

Step 2: Configure the Otto AI Agent

Purpose:
The AI Agent reads the incoming Slack message, determines user intent, and provides structured output to drive the rest of the workflow.

Fields Breakdown:

• AI Agent Model: GPT-4o (Recommended for reasoning and reliability)
• User Query (Prompt): Provide a clear prompt to guide the AI’s response.
  Example Prompt:
  A new message has been received in the Slack channel. {map text field from slack trigger}

Your responsibilities:
1. Analyze the message and classify it as a Bug, Task, or User Story.
2. Create a Trello card based on the type, using the correct format:

Bug Format:
- Clear title
- Steps to Reproduce
- Expected vs. Actual Result
- Impact (e.g., user-facing, blocking)

Task Format:
- Concise title and description
- Background/context
- Assignee and due date (if known)

User Story Format:
- Title: “As a user, I want...”
- Description: Persona, Need, Value
- Acceptance Criteria

Note: Refine and structure the message for clarity and professionalism.
After creating the task, reply in the same Slack thread with a confirmation message and Trello link.
• Session Key: Unique identifier for each user (e.g., Slack User ID) to maintain context across messages.
• Additional Instructions (Optional): Add clarifications such as “Do not assume anything. Ask if information is missing.”
• Fallback Message: Shown if the AI encounters an error (e.g., “Sorry, something went wrong. Please try again later.”)
• Max Runs per Message: Limit the number of AI runs per message (e.g., 3). Leave empty for unlimited.
• Fallback When Max Runs Hit: Message shown if max attempts are reached.

Step 3: Dynamic Mapping — Let AI Decide at Runtime

Unlike traditional workflows that use static field mapping, Otto AI Agents leverage Dynamic Mapping to intelligently populate values at runtime.

What Is Dynamic Mapping?

It’s a flexible, AI-generated output that dynamically fills in fields such as task title and description based on the user message.

Example Dynamic Output:

status: success  
message: Agent configured successfully  
dynamic mapping: AI-generated placeholders  
response output: Final structured message

How to Use It:

In your source action (e.g., Trello – Create Card), map these fields:

• Card Title: {{dynamic mapping}}
• Card Description: {{dynamic mapping}}

This allows the AI to intelligently extract the right content and avoid directly copying raw input.

Step 4: Add Source Actions

What Are Source Actions?

These are the actions executed based on the AI’s output.

Required Source Actions for This Workflow:

1. Trello – Create Card
   • Choose your Board and List
   • Map the following using Dynamic Mapping:
     • Card Name: {{dynamic mapping}}
     • Description: {{dynamic mapping}}
2. Slack – Send Message to Channel
   • Select your Slack channel
   • In Message Text, use {{dynamic mapping}}
   • Set Send as Bot to Yes
   • Customize your bot’s name and icon
   • Map Thread ID using Channel Last Read value from the Slack trigger. This ensures your reply appears in the original message thread.

Testing the Workflow

Try sending this message in Slack:

Hey team – quick heads up  
The Submit button on the Contact Us page isn’t working. I filled in all the fields and clicked submit, but it just shows the spinner and nothing happens. No success message or anything.

Checked the console – looks like a 500 error from the API:  
POST https://api.example.com/contact failed with 500  

Can someone take a look? This seems like a blocker.

Expected Outcome:

• A Trello card is automatically created with structured details.
• A confirmation message is posted in the same Slack thread, with a Trello link.

You can view the full run and responses on the History page.

Final Thoughts

By following this guide, you’ve built a smart, AI-powered workflow that can:

• Understand natural language inputs
• Take intelligent, context-aware actions
• Respond to users—automatically and professionally

Otto AI Agents transform your workflows from static and rigid into adaptive and intelligent automations.