Aircall’s AI Messaging Agents are conversational AI that autonomously handle inbound customer messages on SMS and WhatsApp. They qualify leads, answer frequently asked questions, and automate support, all without scripted decision trees. This article explains what the AI Messaging Agent does, how to configure it, and how to activate it

Key benefits

BenefitDescription
Always-on messaging coverageResponds to inbound messages 24/7, so your team never misses a lead or leaves a customer waiting, even outside business hours.
Automated lead qualification and FAQ handlingCaptures contact information, qualifies leads through intake questions, and answers common queries using your knowledge base.
Intelligent follow-up on outbound workflowsWhen recipients reply to outbound SMS workflows (via HubSpot, the Aircall API, or other tools), the agent continues the conversation on the same line.
Seamless human handoffFor complex queries or when a contact requests it, the agent acknowledges the need for human support and enables your team to follow up with full conversation context.
Shared context across voice and messagingShares the same knowledge base as your AI Voice Agent, ensuring consistent responses across channels, with all AI conversations visible in Aircall Workspace.

How the AI Messaging Agent works

The AI Messaging Agent is a separate product from the AI Voice Agent, with its own tab in the Aircall Dashboard. It handles inbound messages only: it is triggered when a customer sends a message to an Aircall number where the agent is active.

Conversations remain active for 30 minutes of idle time from the last reply. If a customer resumes after that window, the agent retains the conversation context.

The agent works with Aircall SMS and WhatsApp. It can be assigned to the same numbers as your AI Voice Agent or to different lines. When the agent is paused, incoming messages route to your human team through the existing call distribution settings for that number.

Supported channels

ChannelSupported
SMSYes
WhatsAppYes (inbound messages within the 24-hour session window)

Configuration

To configure your AI Messaging Agent, go to Aircall Dashboard > AI Agents > AI Messaging Agent and select your existing AI Messaging Agent or click on Create Agent to open the settings view.

Agent profile

Assign the agent a name. This identifies the agent within the Aircall Dashboard.

Language settings

Select which languages the agent responds in.

Image showing languages dropdown

The agent adapts automatically to the language the customer uses, adjusting spelling and tone accordingly. It can also switch languages mid-conversation: if a customer starts in English and then writes in French, the agent follows.

You must also set a fallback language. This determines how the agent responds when a customer writes in a language that is not among your configured options.

Agent goal

The agent goal tells the agent who it is, what it does, and how it should behave. It sets the guardrails for the agent's responses, following the same principles as Conversational guidelines in the AI Voice Agent.

 
Tip: Keep the agent goal concise. Testing has shown that dense, heavily detailed prompts can cause inconsistent behavior. A short, clearly scoped goal performs better.

Company information

Include a clear description of your company and what the agent is there to help with. A detailed description helps the agent classify customer topics accurately and stay focused on relevant queries.

Knowledge base

The knowledge base is the content the agent draws on to answer customer questions. It works the same way as the knowledge base for the AI Voice Agent, and sources can be shared across both agents.

If a rule or piece of information applies to one channel only, create a dedicated knowledge source for it rather than adding it to a shared source. This prevents the agent from applying channel-specific content in the wrong context.

For guidance on setting up knowledge sources, see Using knowledge sources for your AI Voice Agent.

Messaging script

The messaging script controls what the agent says at three key moments in a conversation:

Script partWhen it is sent
Greeting messageWhen the customer first reaches out. Keep it broad for a generic agent; make it specific if the agent has a dedicated role.
Farewell messageWhen the agent closes the conversation.
Escalation messageWhen the conversation is handed off to a human agent.
 
Important: Escalation and handoff functionality is still a work in progress. When the agent detects an escalation trigger, it stops responding by design to allow a human agent to take over. The escalation message does trigger a handoff, but the end-to-end experience is not yet fully consistent. Test this flow before relying on it in production.

Intake flow

The intake flow defines the information the agent collects from the customer. Each item is configured as a property with two fields:

  • Label: A short identifying name for the property.
  • Description: Context that tells the agent what information to collect. A brief description is sufficient; the agent does not require highly detailed instructions to ask appropriate questions.

You can control how properties are collected:

  • Upfront: The agent asks for all properties immediately after the greeting.
  • Naturally: Properties are gathered through the course of the conversation, based on context.
Image showing enable collecting upfront toggle
Note: AI Actions are available for the AI Messaging Agent and work the same way as in the AI Voice Agent, allowing the agent to trigger actions in connected tools during a conversation. For more details, see Using AI Actions with AI Voice Agents.

Escalation rules

Specify when and how conversations are handed off to a human agent. This is configured alongside the escalation message in the messaging script.

Line assignment

Select which SMS and WhatsApp numbers the agent is active on. The agent can be assigned to the same numbers as your AI Voice Agent or to multiple different lines simultaneously.

 
Important: The AI Messaging Agent can only be assigned to SMS or WhatsApp enabled lines. Voice-only numbers are not supported.

Activation and scheduling

To activate the agent, you need to assign it to a phone number:

Steps:

  1. Go to Aircall Dashboard > AI Agents > AI Messaging Agents.
  2. Select your AI Messaging Agent.
  3. Click on Connect number to assign it.

    Image showing where to find the connect number button
  4. Once assigned, the agent begins responding to incoming messages automatically.

You can configure when the agent is active:

  • Always on: The agent responds to all incoming messages at all times.
  • Scheduled: The agent is active only during specified hours, configurable by time zone and day of the week.

    Image showing schedule options

To deactivate temporarily, pause the agent. While paused, incoming messages route to your human team through the normal call distribution for that number. You can pause the agent in two ways:

  1. From the AI Messaging Agents view, by clicking on the Pause button in the Quick action column.

    Image showing the pause button under the quick action column
  2. In your AI Messaging Agent settings, by clicking on your AI Messaging Agent's name and selecting the Pause agent option.
Pause agent option in the settings view.
Note: If your account is on a free trial, you can create and configure an AI Messaging Agent and test it in the dashboard, but you cannot activate it on a live line. To activate the agent and start handling real customer messages, you will need to upgrade your plan.

Analytics

Basic usage statistics for the AI Messaging Agent are available in the Analytics section of the Aircall Overview and Overview+ Dashboards.

Availability and pricing

  • The AI Messaging Agent is available to all Aircall customers. It does not require an add-on.
  • Pricing will be based on conversations.
  • Pay-as-you-go (pre-paid bundles coming soon)
Monthly conversationsPer conversation
0–1,000$0.35 / €0.30 / £0.26 / AUD $0.53
1,000+$0.29 / €0.25 / £0.22 / AUD $0.44

FREE: 100 conversations to get started + unlimited testing

  • Conversation ends after 30 minutes of idle time from last reply.
  • Standard SMS and WhatsApp rates apply for outbound SMS and WhatsApp templates.

Compliance

You are responsible for ensuring that the phone numbers you connect to the AI Messaging Agent comply with local regulations for automated messaging in each market.

Important: Not all number types are authorised for automated messaging in every country. If you operate in markets with specific regulations around automated or bulk messaging, verify number compatibility before activating the agent on those lines.

Best practices

Use separate knowledge sources for channel-specific content

If content applies only to messaging or only to voice, keep it in a dedicated source rather than a shared one. Use the knowledge base as a source of truth, not an extension of the agent goal. Keep your agent goal focused on role and behavior; put factual content, FAQs, and product details in the knowledge base where the agent can retrieve them accurately.

For best results, ensure your knowledge base content is written in the same language as the customers you expect to serve on that line.

Limit intake properties

Collect only what is genuinely necessary. Fewer questions keep conversations shorter and reduce friction.