Quickstart

This guide will help you create projects, set up workflows, configure features, and deploy your AI companions. Let’s get started!

​

Project Creation and Workflow Setup

​

1. Create a New Project and Workflow

• Navigate to Projects: After logging in, go to Projects tab.
• Create New Project: Click on Create New Project.
• Create Your First Workflow: Provide a workflow name and proceed.
• Choose the modality: Choose your project’s focus, such as voice interactions or chat support.
• Text: Ideal for SMS, chatbots, or other text-based interactions.
• Voice: Perfect for voice-based workflows using calls or phone calls.
• Choose the workflow type: Pick the workflow type that suits your needs
• Simple Workflow: Great for basic interactions without complex logic—like FAQs or simple guidance.
• Advanced Workflow: Offers customization for complex scenarios with multi-step conversations and integrations.

• Create New Project
• Create Your First Workflow
• Choose the modality
• Choose the workflow type

​

Simple Workflow Configuration

Even simple workflows can be powerful! You can enhance them with custom instructions, function calls, and knowledge base integration.

​

1. Configure Your Simple Workflow

• Greeting Message: Set the initial message your users will see.
• Prompt: Provide instructions or context for the companion, guiding its behavior and responses during a conversation.
• LLM Configuration: Choose the close-sources or open-source LLM that suits your needs to power the conversation.
• Voice Providers: For voice mode workflows, choose the voice provider and the associated voice that will handle voice interaction.
• Controls: Advanced settings like temperature allow fine-tuning of how creative or deterministic the AI’s responses should be.

​

2. Add Advanced Functionality

• Function Calls: You can add functions to enhance the capabilities of your AI companion. These function calls can be attached to your prompts, allowing the workflow to pull in real-time data, manipulate information, or trigger external actions based on the user’s response.
  • Custom Functions: Retrieve data from external systems like inventory checks or CRM queries.
  • Pre-defined Functions: Use built-in functions for common actions like ending a call.
• Knowledge Base Integration:
  • Create a Knowledge Base: Navigate to ‘Knowledge’ tab from your dashboard and upload relevant documents to your customer use cases.
  • Enable RAG: Turn on RAG to enable your AI companion to use the knowledge base for more accurate responses. This is useful for answering FAQs, retrieving product documentation, or responding to user-specific inquiries.

• Function Calls
• Knowledge Base Integration

​

3. Share and Test

• Real-Time Testing: Use the Test Chatbot or Test Voice Companion available on the workflow settings sidebar to see your AI in action and make improvements to your workflows based on test results.
• Gather Feedback: A key feature in the Workflow Navigation is the ability to generate a Sharable Link. This allows you to send a link to external stakeholders for testing and feedback, even if they do not have direct access to the platform

• Widget Testing
• Sharable Link

​

Advanced Workflow Configuration

Ready to dive deeper? Upon creating advanced text or voice mode workflows access the drag-and-drop builder interface to visually organize your workflows. Each node represents an interaction or action point within the workflow. You can link these nodes together using conditions to control the flow from one to another, making the workflow dynamic and capable of handling more sophisticated conversational patterns.

​

1. Set Up Your Workflow

• Drag and Drop: Use the visual interface to add different nodes to your workflow canvas.
• Connect Nodes: Link nodes to establish the conversation flow based on conditions or user inputs.
• Configure Nodes: Click on any node to set up prompts, functions, or advanced settings in the sidebar.

​

2. Configure Prompt Nodes

In an advanced workflow, Prompt Nodes are where the conversation between the user and the LLM takes place. These nodes are highly customizable and allow the AI to handle user queries based on predefined instructions or dynamic data pulled from external sources.

• Node Label and Prompt Configuration: Each prompt node can be labeled and configured with specific instructions for the AI. The prompt can include initial greetings, instructions, or context to guide the conversation.
• LLM Configuration: Choose the LLM to power the conversation. Advanced settings like temperature allow fine-tuning of how creative or deterministic the AI’s responses should be.
• Function Calls and Knowledge Base: Similar to simple workflows, you can augment the node’s capabilities with Function Calls (e.g., to pull data or trigger an action) or Knowledge Base Integration (for real-time information retrieval).
• Node Testing: The prompt node can be tested directly in the sidebar, allowing you to validate how it interacts with users before deploying it live.

• Node Label and Prompt Configuration
• LLM Configuration
• Function Calls
• Knowledge Base

​

3. Define Conditions

A key feature of advanced workflows is the ability to define conditions that dictate how the workflow progresses from one node to another:

• Set Conditions: Define rules based on user inputs or variables to direct the conversation path.
• Example: If a user says “Yes,” proceed to the next step; if “No,” redirect them accordingly.

​

4. Workflow Features

At the workflow level, there are several important tools that users can access to manage and test their configurations:

• Workflow Testing: Simulate the entire workflow using the in-builder testing capability, accessible from the workflow settings tab or expanded workflow navigation. This allows you to validate how the workflow responds to different inputs and make adjustments before it goes live.
• Workflow Settings: Configure yoru workflows by setting general system instructions that control the tone and behavior of the AI for the entire workflow.
• Shareable Link: Generate a sharable Link to allow external stakeholders to test the workflow without needing access to the platform. This is especially useful for gathering feedback from clients, colleagues, or other collaborators.
• Workflow Publishing: Once testing is complete, workflows can be published across different deployment channels, such as web chat widgets, SMS, or voice.

• Workflow Testing
• Shareable Link
• Workflow Publishing

​

Conversation Logs

Understanding past interactions helps you improve your AI companion. Access detailed transcripts and technical logs to gain insights in the ‘Logs’ tab available from workflow sidebar navigation.

​

View Transcripts, Technical Logs and Events

• Conversation List: See a list of past interactions with timestamps and user identifiers.
• Detailed View: Click on a conversation to read the full exchange between the user and your AI.
• Search and Filter: Easily find specific conversations by searching or applying filters.
• API Calls: Review external requests made during conversations.
• Response Times: Monitor how quickly your AI responds to users.
• Error Tracking: Identify and troubleshoot any issues that occurred.

​

Deployments

Choose how and where your AI companions are presented and utilized across various channels. The platform provides flexibility to select the best deployment type for different user interactions, customize the interface, and easily integrate it with existing systems.

​

1. Select Deployment Type

Pick the best option for your needs:

• Web Chat Widget: Embed a chat on your website.
• Text iFrame: Integrate using a responsive text frame.
• Voice Widgets: Offer voice interactions within web applications.
• Phone Calls: Automate inbound or outbound calls.
• More Coming Soon: Stay tuned for SMS, Slack, and other channels!

​

2. Customize Deployment

Make it your own:

• Session Triggers: Connect your deployment to an existing workflow to decide when and how the AI initiates the interactions.
• Appearance Settings: Match colors, logos, and styles to your brand.
• Greeting Messages: Personalize how your AI introduces itself to users.

​

3. Integrate with Your Platform

Share the deployment details with your development team:

• Embed Code: Copy the provided code snippet to embed the AI into your site or app.
• API Instructions: For advanced functionalities, follow the API integration guidelines.

​

Inbound/Outbound Deployments

Inbound and outbound deployments define how interactions occur with the assistant, either by the user calling the assistant or the assistant calling the user. To activate a phone number, select a workflow to be linked with that number on this deployment. After selecting and saving the workflow and the phone number, you can call the number to interact with the assigned workflow.

​

1. Inbound Deployment

In an inbound deployment, the user initiates the call to the assistant. The assistant will respond based on the selected workflow.

​

2. Outbound Deployment

In an outbound deployment, the assistant calls the user. You can trigger this call through an API call, where the destination phone number is provided in the payload. From the builder, you can manually input the phone number and select the workflow. When testing from the builder, if you enter your number, the assistant will call you, and the selected workflow will run.