This article answers the most common questions about AI Messaging Agents in Aircall. For setup instructions, see the AI Messaging Agent overview.
General
What is an AI Messaging Agent?
AI Messaging Agents are conversational AI designed to handle inbound customer messages on SMS and WhatsApp. They can qualify leads, answer common questions using your knowledge base, collect information through intake questions, and hand off conversations to a human agent when needed.
Which channels do AI Messaging Agents support?
AI Messaging Agents support SMS and WhatsApp, handling inbound messages on any phone line where they are activated.
Note: Using WhatsApp requires the “WhatsApp in Aircall” add-on.
Are AI Messaging Agents the same product as AI Voice Agents?
They are part of the same AI product family and share the same knowledge base and AI Actions. The underlying technology is optimized for each channel, but from a user perspective, they function as a unified AI product across voice and messaging.
Can I use AI Messaging Agents without AI Voice Agents?
Yes. AI Messaging Agents can be activated independently. You do not need to enable AI Voice Agents to use them.
Do AI Messaging Agents require a specific add-on?
No. AI Messaging Agents are available as a standalone product for all Aircall customers.
Note: The “WhatsApp in Aircall” add-on is required to use WhatsApp.
Setup and configuration
How do I set up the AI Messaging Agent?
Steps:
- Go to the AI Messaging Agents tab in the Aircall Dashboard.
- Create a new AI Messaging Agent.
- Upload your knowledge base content.
- Configure the agent’s objective and intake flow.
- Assign the agent to one or more SMS or WhatsApp lines.
Setup typically takes less than 45 minutes.
Can I use the same knowledge base for AI Messaging and AI Voice Agents?
Yes. Both agents use the same knowledge base. Any updates apply across both channels.
What should I include in the knowledge base?
Focus on frequently asked customer questions. Structure content in a clear Q&A format so the AI can retrieve accurate answers.
Tip: Keep the knowledge base concise and focused. Avoid special characters, as they may confuse the AI.
How many intake questions should I configure?
Limit the intake flow to 3 to 4 questions. This helps keep conversations short while still capturing essential information.
Can I customize the messages the AI sends?
Yes. Use the Messaging script setting to customize greetings, closing messages, and escalation messages to match your brand voice.
Channels and distribution
How does the AI Messaging Agent work on WhatsApp?
The AI responds to inbound WhatsApp messages within the standard 24-hour session window using custom messages, not templates.
Important: If the 24-hour session expires, a new inbound message from the customer is required to start a new session.
Can the AI Messaging Agent initiate conversations?
No. AI Messaging Agents handle inbound messages only. They do not initiate conversations or send WhatsApp template messages.
Can I use the AI Messaging Agent with outbound SMS workflows?
Yes. This is a core use case.
If you send outbound SMS via HubSpot workflows or the Aircall API, the AI Messaging Agent can be activated on the same line to handle incoming replies. It complements existing workflows rather than replacing them.
Is the AI Messaging Agent active 24/7?
At launch, the AI responds to all inbound messages on activated lines around the clock.
Note: Business-hours-only activation is planned for a future release.
AI quality and behavior
How accurate are the AI’s responses?
The AI answers based on your knowledge base. Response quality depends on the accuracy and completeness of that content.
What happens when the AI cannot answer a question?
If the AI lacks sufficient information or the customer requests a human, the conversation is escalated.
Note: The full conversation history and collected data are passed to the human agent, so customers do not need to repeat themselves.
Can the AI make mistakes?
Yes. Like any AI system, output quality depends on the source material.
Important: Regularly monitor AI conversations, especially after activation. Update your knowledge base to address recurring gaps.
How fast does the AI respond?
The AI typically responds within a few seconds. Messaging channels allow slightly longer response times than voice, enabling deeper knowledge base searches when needed.
Human handoff and agent experience
How do agents view AI-handled conversations?
AI conversations appear in Aircall Workspace, where agents can view the full message history.
Note: Improvements to AI conversation visibility in Workspace and mobile apps are in progress.
Can an agent take over an AI conversation at any time?
Not currently.
Note: This capability is planned for a future release, allowing agents to take over, stop, and resume AI conversations.
Can the AI resume after a human agent intervenes?
At launch, once a human takes over, the conversation remains with the human agent.
Note: Resuming AI handling after human intervention is planned for a future release.
Troubleshooting
The AI is not responding to messages. What should I check?
Verify the following:
- The AI Messaging Agent is enabled and active on the phone line
- The line has SMS or WhatsApp enabled
The AI escalates too often. How can I improve this?
This usually indicates gaps in your knowledge base.
Steps:
- Review escalated conversations.
- Identify recurring unanswered questions.
- Update or expand your knowledge base.
- Refine the agent’s objective and prompt.
Can I test the AI Messaging Agent before going live?
Yes.
Steps:
- Use the testing panel in the dashboard.
- Test with a dedicated phone line if needed.
- Send sample messages.
- Verify responses, knowledge base coverage, and escalation behavior.
Compliance
Are there specific compliance requirements?
Yes. Customers are responsible for complying with local regulations in each region.
For example, in the United States:
- Business messaging must be registered under A2P 10DLC
- You must select a use case such as Customer Care, Marketing, or 2FA
- AI-generated content must remain within the registered use case
Important: Many regulations and carrier guidelines require AI systems to clearly identify themselves as an “AI assistant” or “automated system” at the start of the conversation.
Standard opt-in and opt-out rules, such as STOP keywords, continue to apply.