With the customer service AI agent built into 3CX, some routine calls can be handled with no waiting time. Yet the quality of service comes down to one precise moment: when the AI hands over to a person. Here is how 3CX organises this transfer, and what you need to define so that it serves both your customers and your teams.

Good support knows when a person is needed

Artificial intelligence is arriving in phone systems, and with it a legitimate question for any SME: what happens when the AI cannot answer? For Ruth Elizabeth Abbott, COO of 3CX, the answer is an integral part of the solution.

In her view, good support also means knowing when a person is needed: the AI handles what it can, and your teams take over when required.

In other words, the aim is not to replace your support team, but to give it the requests that genuinely call for its expertise. The 3CX customer service AI agent is designed with this in mind: it identifies the request, looks for the answer in the knowledge sources you have assigned to it and, when necessary, transfers the call.

Two situations trigger a transfer

The request falls outside the AI's scope

A question that is too complex, a special case, a topic not covered in the knowledge sources: the agent must not improvise. The call is then transferred to the transfer destination you have configured.

The caller asks to speak to a person

Some customers simply want to talk to a human. The agent respects this request and transfers the call, without trying to keep the caller in a conversation they do not want.

In most deployments, the default destination is a support queue. To go further, 3CX offers escalation topics: a request relating to billing, a technical problem or a particular specialism can be directed to a user, a queue or another, more specialised customer service AI agent.

Picking up the call with full context

A successful transfer is about more than ringing the right extension. Nothing annoys a customer more than having to repeat their problem. With 3CX, the team can view the transcript of the conversation between the caller and the AI agent: the person taking over the call knows what was asked, what has already been answered and where the conversation stopped.

For your staff, this is an immediate time saver. For your customers, it is one continuous journey, not two conversations that start from scratch.

What to define before going live

The transfer needs to be prepared in advance. Before activating the agent, we recommend answering a few simple questions:

  • Scope: which requests should the agent handle, and which should go straight to your teams?
  • Default destination: to which queue or group are unresolved calls sent?
  • Escalation topics: which subjects warrant a specific destination (billing, technical, sales)?
  • Organisation: is the chosen destination consistent with your opening hours and your phone system's routing rules?
  • Testing: has the agent been tested with realistic requests before being exposed to your customers?

3CX also advises deploying several targeted agents rather than a single agent expected to cover every scenario. Each one then has its own knowledge sources, greeting, transfer destination and escalation topics.

Technical prerequisites

The customer service AI agent requires the 3CX AI edition, version V20 Update 10 or later, a supported real-time API provider, prepared knowledge sources and a destination for unresolved calls.

Stay in control of the customer journey

This division of roles between AI and humans is nothing new for us. It is the approach we apply with our partners: a first level that handles routine requests, a second level that takes care of complex cases, and a clearly defined escalation process between the two. The 3CX AI agent fits the same model: it becomes a first point of contact, provided the transfer rules are designed for your organisation.

At MLS Communications, we help you define this scope, configure destinations and escalation topics, then fine-tune everything based on feedback from your teams. Whether you want to stay in control, delegate or combine both, you choose your level of autonomy with our support services.

Would you like to add an AI agent to your 3CX phone system? Contact us to analyse your call flows and define together what the AI handles and what goes to your teams.