Channels

The Channels settings page is the first under Configuration and lists all configured channels, allowing you to add or modify how ept AI connects to external systems.

Overview

Channels connect the AI to external environments such as forums, support systems, or chat platforms. Each channel uses a Knowledge Source Configuration (KSC) that defines which knowledge sources the AI may retrieve from.

Before You Go Live

Confidentiality Cascade Rules: If any source in a KSC is confidential → the KSC is confidential → only confidential channels can use it.

See detailed explanation: Confidentiality Rules

Key Capabilities

Channel Management

Create Channel

The "Create Channel" button opens a wizard to add a new channel:

Required Information:

• Channel Name: Descriptive name for the channel
• Company Name: Optional text field to specify the company associated with the channel. This helps the AI tailor responses to the specific client or business unit. Recommended when the channel serves a particular customer or organization.
• Channel Type: Select from supported integration types (see complete list below)
• Knowledge Source Configuration: Choose which KSC the channel should use (required)
• Confidentiality Toggle: Set whether channel can access confidential content
• Connection Details: Provide required credentials and integration settings

Channel Type Options: The Select Type dropdown includes the following options:

• Community Forum: Integration with public community forums and discussion boards
• Chat: Direct chat interfaces and embedded chat widgets
• Email: Integration with email systems for AI-assisted responses
• HubSpot: Integration with HubSpot CRM and support systems (see detailed setup below)
• Zendesk: Integration with Zendesk ticketing and support systems
• Slack: Integration with Slack workspaces for team AI assistance
• Websocket: Custom websocket integration for real-time messaging applications
• ept AI App: Native ept AI chat interface and deal room channels

Administrator Only

Only administrators should create or modify channels, as incorrect configuration can affect AI functionality across the organization.

Channel Table

The main table displays existing channels:

Column	Description
Name	Descriptive name for the channel
Type	Integration type (see supported channel types below)
Confidential	Checkbox indicating if channel can access confidential content
Knowledge Source Configurations	Associated KSC that provides content access

Supported Channel Types

Communication Platforms

AI Chat (Web/Mobile)

Direct chat interface accessible via web browser or mobile app. Provides the primary user interface for direct AI interaction with full conversation history and context.

Community Forum

Integration with public community forums and discussion boards. ept AI can monitor forum posts and provide suggested responses to moderators, or automatically respond to questions based on configured knowledge sources.

Use Cases:

• Automated responses to common questions in community forums
• AI assistance for forum moderators and community managers
• Consistent information sharing across forum discussions

Email

Integration with email systems for AI-assisted email responses. ept AI can analyze incoming emails and generate suggested responses that users can review and send.

Use Cases:

• Customer support email assistance
• Sales inquiry response suggestions
• Internal email automation for common requests

Slack

Integration with Slack workspaces for team AI assistance within channels. Team members can query the AI directly within Slack channels or receive AI-powered insights.

Configuration Requirements:

• Slack app installation and permissions
• Workspace integration setup
• Channel-specific configuration for AI access
• Bot token and API credentials

Use Cases:

• Technical support queries within development teams
• Product information lookup during sales conversations
• Internal knowledge sharing and documentation access

WebSocket

Custom websocket integration for real-time messaging applications. Enables embedding AI chat functionality directly into external applications with real-time bidirectional communication.

Configuration Requirements:

• Websocket endpoint configuration
• Authentication and security setup
• Message format and protocol definition
• Real-time connection management

CRM and Sales Platforms

HubSpot

Direct integration with HubSpot CRM and support systems. Provides AI-generated responses that appear as Notes in HubSpot contacts and tickets. See the comprehensive HubSpot Channel Integration section for detailed setup instructions.

Salesforce

Integration with Salesforce for deal and customer context. Enables AI-assisted responses within Salesforce workflows and customer interactions.

Deal Room

Specialized channels for individual sales opportunities with full CRM context. Each deal room creates its own channel with access to deal-specific information and stakeholder data.

Support Systems

Zendesk

Integration with Zendesk ticketing and support systems. ept AI automatically analyzes new tickets and generates suggested responses for support agents.

Setup Process:

1. Install the ept AI Zendesk app from the Zendesk marketplace
2. Configure API credentials and permissions
3. Set up ticket monitoring and response triggers
4. Configure knowledge source access and confidentiality settings
5. Test the integration with sample tickets

Functionality:

• Automatic analysis of new support tickets
• AI-generated response suggestions appear as internal notes
• Integration with Zendesk's ticket workflow and status updates
• Support for multi-language tickets and responses

Best Practices:

• Ensure knowledge sources include current support documentation
• Configure appropriate confidentiality settings for customer data
• Train support agents on reviewing and customizing AI suggestions
• Monitor response quality and adjust knowledge sources as needed

Support Portal

Integration with customer support portals and help desks. Provides AI-powered responses within existing support infrastructure.

Ept AI Channels

Ept AI App

Native ept AI chat interface and deal room channels. Includes all standard ept AI functionality with full feature access and advanced conversation management.

Integration Requirements

Each channel type has specific configuration requirements:

API-Based Integrations (Slack, Teams, etc.):

• API keys or OAuth tokens
• Webhook URLs for real-time communication
• Permission scopes for accessing relevant data

CRM Integrations:

• CRM credentials and API access
• Field mapping configuration
• Data synchronization settings

Forum/Support Integrations:

• Forum API credentials
• Content filtering rules
• Response posting permissions

Channel Actions and Interactive Features

Table Management

• Search Functionality: Search channels by name, type, or configuration details
• Column Sorting: Sort by Name, Type, Confidential status, or KSC
• Filter Options: Filter by channel type, confidentiality status, or connection status
• Pagination: Navigate through large channel lists efficiently

Row Actions

• Edit: Click any row to open the channel detail page
• Status Monitoring: View channel connection status and health indicators
• Configuration Review: Check settings, credentials, and integration status
• Actions Menu: Vertical-ellipsis menu with additional options:
  • Duplicate Channel: Create a copy of the channel configuration
  • Test Connection: Verify channel connectivity and functionality
  • View Logs: Access connection and performance logs
  • Export Configuration: Download channel settings for backup
  • Delete Channel: Remove channel (admin only)

HubSpot Channel Integration

The HubSpot channel integration allows ept AI to respond directly within HubSpot for both sales and support workflows, automatically generating AI-powered responses that appear as Notes in HubSpot.

Use Cases

Sales Use Case

When the custom contact property ept_ai_message is updated for a contact, ept AI automatically posts a Note in the contact's activity timeline with a suggested response. Sales representatives can copy this AI-generated text when replying to prospects or following up on leads.

Workflow:

1. Lead submits a form or sends a message
2. HubSpot workflow copies the message to the ept_ai_message property
3. ept AI detects the property change and generates a response
4. AI response appears as a Note in the contact's timeline
5. Sales rep copies and uses the suggested response

Support Use Case

When a support ticket is created in HubSpot, ept AI generates a reply and attaches it as a Note on the ticket. Support agents can use the suggested answer as a starting point for their response, ensuring consistency and leveraging the AI's knowledge base.

Workflow:

1. Customer creates a support ticket
2. ept AI analyzes the ticket content
3. AI generates a suggested response based on knowledge sources
4. Suggested response appears as a Note on the ticket
5. Support agent reviews and uses the AI-generated response

Setup Instructions

Step 1: Install the ept AI HubSpot App

1. Contact ept AI support to receive the special installation link for the HubSpot app
2. Click the installation link and approve the app permissions in your HubSpot account
3. Verify the app appears in your HubSpot Connected Apps section

Step 2: Create Custom Contact Property

1. In HubSpot, navigate to Settings > Properties > Contact Properties
2. Create a new property with these specifications:
   • Property Name: ept_ai_message
   • Object Type: Contact
   • Group: Conversion Information
   • Field Type: Multi-line text
   • Description: "Message content for ept AI processing"

Step 3: Configure HubSpot Workflows

1. Create or update your HubSpot forms to include a message field
2. Set up a HubSpot workflow that triggers when:
   • A form is submitted, or
   • A ticket is created, or
   • Another relevant trigger occurs
3. Configure the workflow action to copy the form's message field into the ept_ai_message property
4. Ensure the workflow is active and properly configured

Step 4: Create the ept AI Channel

1. In ept AI, navigate to Configuration > Channels
2. Click "Create Channel" and select "HubSpot" as the channel type
3. Provide the required HubSpot connection details and credentials
4. Select the appropriate Knowledge Source Configuration
5. Configure confidentiality settings based on your security requirements
6. Save the channel configuration

Step 5: Test the Integration

1. Submit a test form or create a test ticket in HubSpot
2. Verify that the ept_ai_message property is updated correctly
3. Check that an ept AI-generated Note appears within approximately one minute
4. Review the Note content for accuracy and relevance
5. Contact support if the integration doesn't work as expected

Data Usage and Privacy

Data Access: ept AI accesses only the following HubSpot data:

• Ticket body content (for support use cases)
• The ept_ai_message property value (for sales use cases)
• Basic contact information required for context

Data Security:

• No other HubSpot data is accessed or processed by ept AI
• All data transmission is encrypted and secure
• Data usage follows ept AI's standard privacy policies
• Contact information is used only for generating relevant responses

Compliance:

• Integration respects HubSpot's data handling policies
• Data processing complies with GDPR and other relevant regulations
• Customer data is processed only as necessary for AI response generation

Troubleshooting

No Notes Appearing

If ept AI Notes are not appearing in HubSpot:

1. Verify Property Updates: Check that the ept_ai_message property is being updated correctly
2. App Authorization: Ensure the ept AI HubSpot app is properly installed and authorized
3. Workflow Configuration: Verify HubSpot workflows are active and correctly configured
4. Channel Status: Check the ept AI channel status and connection health
5. Timing: Allow up to 2 minutes for Notes to appear after property updates
6. Permissions: Verify the HubSpot user has appropriate permissions for the integration

Incorrect or Poor Quality Responses

If AI responses are not relevant or accurate:

1. Knowledge Source Review: Check that the channel's KSC includes appropriate knowledge sources
2. Property Content: Verify the ept_ai_message property contains clear, complete information
3. Channel Configuration: Review channel settings and confidentiality configuration
4. Context Information: Ensure adequate context is provided in the message content

Integration Performance Issues

If the integration is slow or unreliable:

1. HubSpot Status: Check HubSpot system status and API availability
2. Network Connectivity: Verify network connectivity between systems
3. Rate Limits: Ensure API rate limits are not being exceeded
4. Support Contact: Contact ept AI support for performance optimization assistance

Best Practices

For Sales Teams

1. Clear Messaging: Ensure form fields capture clear, complete customer inquiries
2. Context Addition: Include relevant context about the prospect or opportunity
3. Response Review: Always review AI-generated responses before sending to customers
4. Follow-up Tracking: Use HubSpot's activity tracking to monitor response effectiveness

For Support Teams

1. Ticket Detail: Ensure support tickets contain sufficient detail for accurate AI responses
2. Knowledge Base Maintenance: Keep knowledge sources updated with current support information
3. Agent Training: Train support agents on how to effectively use AI-generated suggestions
4. Quality Monitoring: Regularly review AI response quality and provide feedback for improvement

For Administrators

1. Regular Testing: Periodically test the integration to ensure proper functionality
2. Permission Management: Regularly review and update HubSpot app permissions
3. Performance Monitoring: Monitor integration performance and response times
4. User Training: Provide comprehensive training on the integration workflow
5. Security Review: Regularly review data access and security configurations

Channel Detail Page

Clicking a channel row opens a detailed configuration page:

Basic Information

• Channel ID: Read-only unique identifier
• Channel Name: Editable name with pencil icon for renaming
• Channel Type: Read-only type indicator (e.g., ept-ai-app, hubspot)
• URL: Read-only integration endpoint URL

Security Settings

Confidential Toggle

A critical setting that controls content access and must align with the associated KSC:

• When ON: Channel can access confidential Knowledge Source Configurations
• When OFF: Channel can only access non-confidential KSCs
• Automatic Inheritance: Channel inherits confidentiality if its KSC contains any confidential sources
• Access Control: Determines what information the AI may surface in responses

Confidentiality Rules:

1. If any knowledge source in a KSC is confidential, the entire KSC becomes confidential
2. Confidential KSCs can only be used by channels marked as confidential
3. Changing channel confidentiality may require changing the associated KSC
4. Public channels cannot access confidential content, regardless of user permissions

Security Impact

The confidential toggle is automatically enforced based on KSC confidentiality. You cannot assign a confidential KSC to a non-confidential channel. Ensure channel confidentiality matches your security requirements and user access needs.

Knowledge Configuration

Knowledge Source Configuration (KSC) - Required

Every channel must be linked to exactly one Knowledge Source Configuration:

• Current KSC: Displays the currently associated knowledge source configuration
• Change KSC: Pencil icon opens dropdown to select different KSC
• Immediate Effect: Changes take effect immediately and alter which knowledge sources the AI uses
• Confidentiality Inheritance: If the selected KSC contains any confidential sources, the channel must also be marked as confidential

Impact of KSC Changes:

• Affects answer quality and content availability
• May require channel confidentiality adjustment
• Changes which knowledge sources AI can access for responses
• Should be done carefully as it affects all users of the channel

Advanced Actions

Delete Channel

• Trash Icon: Permanently deletes the channel
• Warning: Only administrators should perform this action
• Impact: Deletion affects all applications using this channel

Monitor Responses

• Responses Button: Direct link to responses filtered for this specific channel
• Usage: Audit recent AI answers generated through this channel

How Channels Work

Integration Flow

1. External Request: User interacts with external system (forum, chat, CRM, etc.)
2. Channel Authentication: Channel validates request and user permissions
3. Context Gathering: Channel gathers relevant context (company, user, conversation history)
4. Knowledge Retrieval: AI searches configured knowledge sources via the associated KSC
5. Content Filtering: System applies confidentiality and access controls
6. Response Generation: AI generates response based on retrieved information and context
7. Response Delivery: Answer is returned through the channel to the external system
8. Activity Logging: Interaction is logged for performance management and analysis

Knowledge Access Control

• Mandatory KSC Association: Each channel must link to exactly one KSC
• Confidentiality Enforcement: Channel confidentiality must match or exceed KSC confidentiality
• Source Limitations: AI can only access knowledge sources included in the channel's KSC
• Dynamic Filtering: Access is filtered in real-time based on channel and user permissions
• Context Awareness: Some channels (like Deal Rooms) provide additional context for more targeted responses

Troubleshooting

Channel Not Working

If a channel isn't functioning properly:

1. Credentials Check: Verify integration credentials on the detail page
2. Knowledge Sources: Ensure associated knowledge sources are valid and up-to-date
3. Confidential Setting: Check if confidential checkbox matches security requirements
4. KSC Validation: Verify the KSC contains appropriate knowledge sources
5. Connection Test: Test external system connectivity

Missing Content

If the AI isn't finding expected information:

1. KSC Review: Check which knowledge sources are included in the channel's KSC
2. Confidentiality Mismatch: Verify channel confidential settings align with KSC requirements
3. Knowledge Source Status: Ensure relevant knowledge sources are active and processed
4. KSC-Channel Alignment: Confirm the channel's KSC includes the expected knowledge sources
5. Access Permissions: Verify the channel has appropriate access to necessary content
6. Integration Status: Check that external integrations (CRM, forum, etc.) are functioning properly

Permission Issues

If you can't modify channels:

1. Administrator Rights: Only administrators can edit channel settings
2. Role Verification: Check your user role and permissions
3. Contact Admin: Request assistance from system administrators

Best Practices

For Administrators

1. Security First: Always verify confidential settings match content sensitivity
2. Clear Naming: Use descriptive names that indicate channel purpose and scope
3. Regular Audits: Periodically review channel configurations and usage
4. Documentation: Maintain records of what each channel is used for
5. Testing: Test channels after configuration changes

Channel Configuration Best Practices

When to Mark Channels as Confidential

Mark as Confidential when:

• Channel will access internal company documentation or proprietary information
• Channel serves specific clients with sensitive business data
• Channel includes competitive intelligence or strategic information
• Channel processes customer data that requires privacy protection
• Associated KSC contains any confidential knowledge sources

Keep Non-Confidential when:

• Channel serves public-facing support or general inquiries
• Channel uses only public documentation and marketing materials
• Channel provides general product information to prospects
• Associated KSC contains only public knowledge sources

When to Provide Company Names

Provide Company Name when:

• Channel serves a specific client or customer organization
• AI responses should be tailored to a particular business context
• Channel is used for client-specific support or sales activities
• Company-specific knowledge or customization is required
• Channel integrates with a specific organization's systems (HubSpot, Zendesk, etc.)

Leave Company Name Blank when:

• Channel serves general public inquiries
• Channel provides internal team support across multiple clients
• Channel uses generic knowledge sources without company-specific content
• Channel is used for product demonstrations or general information sharing

Integration-Specific Configuration

HubSpot Channels:

• Always provide company name to align with HubSpot account context
• Mark as confidential if accessing customer-specific data
• Configure appropriate knowledge sources for sales or support use cases

Zendesk Channels:

• Provide company name when serving specific client support
• Mark as confidential when processing customer tickets with sensitive information
• Ensure knowledge sources include current support documentation

Slack/Internal Channels:

• Company name may not be necessary for internal team channels
• Mark as confidential for internal knowledge and proprietary information
• Configure knowledge sources appropriate for internal team needs

For Operations Teams

1. Monitor Performance: Use the Responses button to track channel effectiveness
2. Knowledge Updates: Coordinate with admins when knowledge sources need updates
3. User Training: Ensure users understand which channels to use for different purposes
4. Issue Reporting: Report connectivity or content issues promptly

Integration Planning

1. KSC Strategy: Plan knowledge source configurations before creating channels
2. Security Classification: Clearly define what content should be confidential and plan KSC confidentiality accordingly
3. Channel Mapping: Map external systems to appropriate knowledge configurations based on use cases
4. Confidentiality Architecture: Design channel and KSC relationships to properly control access
5. Scalability: Consider future growth when designing channel architecture
6. Integration Requirements: Gather all necessary credentials and permissions before channel creation
7. User Training: Plan training for users on which channels to use for different purposes

Related Features

• Knowledge Source Configurations - Create and manage the KSCs that channels use
• Knowledge Sources - Manage the content repositories that power channels
• Responses - Monitor AI interactions through specific channels
• Users - Manage who can configure and access different channels