This guide helps your team get more value from TurboHelp's AI and Automation tools. Use it to understand the feature, configure it correctly, and know what to check before relying on it in your daily support workflow.
At a Glance
Use automation to remove repetitive work while keeping humans available for complex issues.
Start with focused rules or flows, test them with real examples, and expand gradually.
Review AI handoffs, tool results, and knowledge quality so automation stays helpful.
AI Agent flows let you design guided conversation paths for common support scenarios. Instead of relying only on free-form AI responses, flows give you structure: ask for details, show options, call tools, branch by customer input, and escalate to a human when needed.
The Flow Canvas
The flow canvas is the visual workspace where you build the customer's path. Each flow starts with a start point and continues through connected steps. Keep each flow focused on a single job, such as checking order status, collecting billing details, or routing a technical issue.
Start Point and Intent
The start point defines when the flow should run. A flow may be launched from an AI Agent greeting, a button, a customer intent, or another flow. Use clear intent descriptions so the AI Agent can choose the right flow for the customer's request.
Common Flow Steps
Message: Send text to the customer, explain what will happen next, or provide an answer.
Buttons: Present predefined choices so the customer can pick a path.
Ask for details: Collect structured information such as email, order number, product, or issue type.
Set attribute: Store information on the user or conversation for routing, reporting, or personalization.
Branches: Create conditional paths based on attributes, customer answers, or tool results.
Use tool: Call an external API or internal service to fetch data or perform an action.
Show articles: Suggest help center articles that may solve the customer's issue immediately.
Transfer: Hand the conversation to a human agent or group when automation should stop.
Close conversation: End the flow when the issue is resolved.
Testing a Flow
Before publishing a flow, test the happy path and the failure path. Make sure customers can recover from missing information, failed tool calls, unclear answers, and requests to speak with a person.
Flow Design Tips
Keep each step short and easy to understand.
Use buttons when the customer should choose from a known set of options.
Use branches to keep logic visible instead of hiding too much inside long instructions.
Always include a transfer path for sensitive, urgent, or unresolved issues.
Review completed conversations to improve flow wording and routing over time.
Best Practices
Give the AI agent narrow, high-quality knowledge before broadening its scope.
Always define a clean escalation path to a human agent.
Use test conversations to confirm the automation does what customers expect, not just what the rule says.