Version: Latest

Flow Builder — Best practices

Structure flows

Structuring flows effectively can make your chatbot development process smoother and more maintainable.

  • Organise Flows by names: Use the same prefix the naming flows that are within the same domain/skill to make it easier to locate and work with flows as your assistant grows. i.e Flow 1: open_account_eligible, Flow 2: open_account_ineligible
  • Prevent Flow Duplications: Before creating a new flow, check if a similar flow already exists (similar description, similar steps). Duplicating flows can lead to maintenance challenges and confusion. Instead, consider reusing existing flows when applicable.
  • Link Flows Where Possible: When designing your bot's conversation paths, look for opportunities to link existing flows together. This is also one way to reuse flows and can reduce redundancy and make your bot's logic more streamlined.
  • Continuously Review and Test Your Bot: After structuring your flows, periodically review and test your assistant to ensure that the conversation paths and logic work as intended or if you need to add/remove/rearrange some of the flows.

Common conversation elements

Including in your assistant flows that cover the common conversation elements can help ensure that your chatbot is well-rounded and capable of handling various user interactions. Here's a checklist of essential flows to consider:

  1. Greeting: Implement a flow with a friendly and context-aware greeting message to welcome users as they initiate a conversation.
  2. Introduction: Create a flow that provides a brief introduction to the bot's purpose, capabilities, and how it can assist the user.
  3. User Help: Include a flow or path that can be triggered when users request help or guidance during the conversation.
  4. Exit or Goodbye: Create an exit flow where you include an exit or goodbye and prompts for feedback when users decide to end the conversation or complete a task.This encourages users to provide feedback on their experience to help improve the bot's performance.
  5. Privacy and Data Handling Information: Create a flow about privacy and data handling and inform users about how their data is handled and assure them of privacy and security measures. Offer users the option to opt-out of the data procession.
  6. FAQ Handling: Create series of flows for FAQs and ensure the assistant can provide relevant information.
  7. Human Handoff: Implement a flow for transferring the conversation to a human agent when the assistant can't handle a user's request effectively.

Create human handoff flow

A human handoff skill is essential for scenarios where your assistant encounters complex or sensitive questions that it can't handle on its own. In this guide, we’ll look at how to add this handoff skill as a backup plan to ensure a smooth transition from assistant to human.

Step 1. Identify Handoff Trigger Scenarios: Determine the specific situations in which your assistant should trigger a human handoff as backup plan. These could include questions about sensitive topics, complex issues, or when the user explicitly requests human assistance

Step 2. Integrate Handoff Logic: Implement the logic that triggers the human handoff skill. This logic should be based on the scenarios you identified in step 1. For example, you can use specific keywords or phrases as triggers.

Step 3. Prepare User for Handoff: Before transferring to a human agent, inform the user that a human is taking over to provide assistance. Set clear expectations about the handoff process and reassure the user.

Step 4. Integrate with Human Support: Ensure that your assistant can seamlessly connect with human support agents. This might involve integrating with a helpdesk system, customer support platform, or routing requests to a designated team.

Step 5. Test and Refine: Thoroughly test the bot's behavior when triggering the handoff. Make adjustments and refining the handoff triggers as needed to improve the handoff experience.