Skip to main content

Understanding AI Agents in Whippy

A high-level guide to what AI agents are, how they work, and how they operate across both messaging and calling.

Maria Cairns avatar
Written by Maria Cairns
Updated over a week ago

Why it matters

AI agents automate conversations and tasks across your communication channels. Understanding how they work helps you design consistent, reliable behavior without needing separate agents for each mode. One agent can handle messages, calls, or both, depending on where it is assigned.

Key Concepts

Agent: A configurable AI assistant that can operate in chat or voice interactions using a single unified set of instructions.

Agent Instructions: A text field where you define the agent’s purpose, behavior, communication style, and channel-specific guidance.

Channels: The surfaces where an agent can operate, such as messaging inboxes or phone numbers. The agent’s capabilities depend on where it is assigned, not how it is created.

Knowledge Base: Optional reference material the agent uses to improve accuracy across both calls and messages.

Agent Settings: Controls that affect voice delivery, transcription, language handling, responsiveness, and other call-related behaviors.

Data Parsing: Optional extraction of structured information from conversations.

Versions: A record of saved changes to your agent.

Step-by-Step: Explore and Understand Agent Configuration

  1. Open the Sparkles icon from the left navigation.

  2. Click Agents to view your agent list.

  3. Select any agent to open the Agent Editor.

  4. Review Basic Info to understand the agent’s name, description, and status.

  5. Open Agent Instructions to see the core rules defining how the agent behaves.

  6. Look for channel-specific guidance, such as how the agent should speak on calls or message in chat.

  7. Review Call Instructions and Agent Settings to understand how calls are handled.

  8. Open Data Parsing and Analysis to see how the agent extracts structured information.

  9. Review Knowledge Bases attached to the agent to understand what content it can reference.

  10. Check Versions to see how the agent has evolved over time.

Tips and Best Practices

  • Remember that agents are not tied to a channel. One agent can be used for messaging, calling, or both.

  • Keep the Instructions field clear and explicit, especially when defining differences between chat and call behavior.

  • Use Knowledge Bases to keep instructions shorter and more maintainable.

  • Review call logs and chat transcripts to understand how the agent interprets your instructions.

  • Use Draft mode while learning or making changes, then switch to Active when stable.

Troubleshooting

Issue

Possible Cause

Fix

Agent behaves differently across modes

Missing channel-specific notes

Add explicit guidance in the Instructions field

Agent cannot take calls

Not assigned to a call channel

Assign to a phone number in Inbound Call Control

Messages seem off-tone

Instructions too vague

Add clearer tone and style expectations

Data fields are missing after calls

Parsing model not configured

Review Data Parsing settings

Did this answer your question?