Integrating AI Voice

AI Voice brings automated, real-time conversational calling into your multi-channel customer journeys. By placing voice calls alongside Email, WhatsApp, and SMS, you can re-engage customers precisely when text-based channels go quiet or when an urgent touchpoint is required.

AI Voice supports both:

• Outbound Calls: Triggered automatically by customer actions or journey delays (e.g., payment recovery, delivery confirmation).
• Inbound Calls: Answering incoming customer queries instantly and escalating to a live human agent with full context when necessary.

Define the Journey Trigger

Identify the point in your Customer Journey where a voice call adds value. Common triggers include:

• No Response Timeout: A user opens a WhatsApp or SMS message about a pending payment or delivery but takes no action within a set time frame (e.g., 2 hours).
• High-Priority Events: Urgent notifications requiring immediate confirmation.

Connect via Webhook

To trigger an AI Voice outbound call or pass contextual user data, insert a Webhook node into your journey flow:

• Add a Webhook Action node directly after your condition/timeout block.
• Configure the HTTP POST request to endpoint targeting the AI Voice service.
• Pass the required user properties in the payload (e.g., phone_number, first_name, order_id, custom_context).

For full specifications on payload formatting, header parameters, response handling, and security signatures, refer to our Webhook Documentation.

Handle Call Outcomes and Agent Handoffs

Once the AI Voice call completes, the system returns status events back to indigitall (e.g., call_completed, payment_confirmed, transferred_to_human).

• Successful Resolution: Continue the customer along the primary journey path.
• Escalation / Human Handoff: If the customer requests live assistance during the call, AI Voice routes the call to your support queue along with the full conversation transcript and customer context.

Best Practices

• Timing Matters: Set reasonable quiet hours (e.g., do not trigger outbound calls between 9:00 PM and 8:00 AM in the recipient's local time zone).
• Fallback Options: If an AI Voice call goes unanswered or to voicemail, route the user to a secondary SMS or email follow-up node within the journey.

FAQ

How does the AI Voice Agent work?

The AI Voice Agent connects to your systems in real time. During a call, it can read information such as the customer's name, their profile, what happened on their last call, available time slots in your calendar, or the status of an order.

It can also write information and take action while the call is in progress. For example, it can:

• Update a customer record
• Book an appointment
• Unsubscribe a customer on request
• Send a confirmation email
• Send a one-time password (OTP) to verify the customer's identity and help prevent fraud
• Send a WhatsApp message
• Transfer the call to a human agent

Can the AI Voice Agent access real-time information during a call?

Yes, as long as you have an API it can query. For example, to check the status of an order, the agent asks the customer for their order number, queries your API, and immediately tells them whether the order is on its way, has been delivered, or is still in the warehouse and will ship in the next few days.

Can the AI Voice Agent transfer a call to a human?

Yes. When the agent can't resolve the customer's issue or question, it can automatically transfer the call to a human agent based on the context of the conversation, for example when it detects that the customer is unhappy.

You can also configure the agent to offer the customer the option to speak with a person. Because the agent uses natural language, the handoff feels seamless. This way, you only route specific calls to your team instead of relying on the AI agent for every case.

What happens if the customer interrupts, speaks fast, or there's background noise?

The AI Voice Agent is designed to handle interruptions: when the customer starts talking, the agent automatically stops speaking and listens. It can also understand the customer in noisy environments and interpret everyday expressions.

In addition, you can configure several parameters, such as the agent's speaking speed (faster or slower), so it's ready to answer any type of question and interact naturally with your customers.

What languages does the AI Voice Agent speak?

You decide which language the agent uses at the start of each call. You can even set it based on a customer attribute. For example, call one customer in English and another in Spanish.

The agent can be trained to speak multiple languages, such as English and Spanish, and to use local variants such as Mexican, Colombian, or Brazilian (distinguishing Brazilian Portuguese from European Portuguese). This way, customers feel they're being served locally, in their own language. You can also use different phone numbers for different languages and countries.

What phone number does the AI Voice Agent call from?

It depends on the country, since each country (for example Spain, Chile, or Colombia) has its own regulations, and indigitall complies with them when assigning numbers. For example, Spain has introduced new regulations that no longer allow commercial calls from mobile numbers, so these calls must come from other number types, such as 900 or geographic numbers.

indigitall can provide phone numbers for you. If you already have your own numbers, you can also import them.

How does indigitall handle security, privacy, and compliance?

• Consent (opt-in): You can only call customers who have agreed to receive outbound calls. You can't buy a phone database and start calling people.
• Opt-out management: If a customer withdraws their consent at any time, they're automatically flagged, and they won't be called again. The agent can also check do-not-call lists (such as the Robinson List in Spain) to exclude customers who have opted out.
• AI and recording disclosure: At the start of each call, the agent tells the customer that they're speaking with an AI agent and that the call will be recorded.
• Regulations: All data is stored in compliance with GDPR and, in the United States, HIPAA.
• Data retention: Call records can be deleted according to the retention period you choose, so you meet your company's legal requirements.

How does the AI Voice Agent integrate with my current systems?

Integration is simple. If you already have a CRM, indigitall syncs it with its Customer Data Platform (CDP). This lets you choose which customers should receive an outbound call and why.

indigitall also provides APIs that create events in real time. Integration can be as simple as uploading a CSV file with the customers you want to call, or as advanced as triggering a journey from an event, such as an upcoming patient appointment or a student who has enrolled or requested more information. That journey then triggers the phone call as one more channel.

How does the AI call fit into the Customer Journey?

indigitall integrates the AI call directly into the Customer Journey, so voice becomes one more channel alongside the rest of your campaigns. The journey decides when to make the call and what happens next based on the result.

Who controls the call?

indigitall's CDP controls the call and stores everything that happens during it. This is what sets indigitall apart: the call is simply one more channel in the journey, and the CDP decides when to move to the next step. For example:

• If the customer doesn't answer, the journey triggers another action.
• If the customer answers, the agent queries and updates your systems in real time.

Nothing gets lost. When the call ends, the recording and, optionally, the transcript can be uploaded directly to the customer's profile (as long as the customer was told the call would be recorded). This gives you a full history of every call with each customer and what happened on it, including proof that a customer requested a specific product.

What reports are available?

The reports show all your call activity: how many calls were made, at what time, and what happened during them. You can view reports by phone number and by country, and see what each call was about.

You can go from a high-level summary (total calls, completed calls, and outcomes) down to call-level detail.

How does the AI Voice Agent improve customer satisfaction (NPS)?

Instead of a traditional "press 1, press 2" menu, the agent talks to customers in their own language, using natural language, and responds quickly. Because it can handle every call in real time, whether it's an inbound call, a "Call me" button request, or an outbound call triggered by an event, customers get their questions answered right away, which raises satisfaction and NPS.

What do I need to set up the AI Voice Agent?

Setting up a pilot is quick. The AI agent is trained on your documentation and connected to your APIs. It can even learn from your current calls, including those handled by human agents.

You can then launch a use case with a single phone number and measure the results: see how the AI agent converts, what return you get from the calls, and how much you save. This makes it easy to evaluate voice as one more channel in your customers' journey.