Skip to content

EXPERIENCE

AI for customer service on voice, WhatsApp and digital channels

AI agents that understand the customer, query your systems and resolve on first contact, with clear limits on autonomy and handover to an agent with the full context.

WHAT WE DO

AI customer service is no longer a menu chatbot. Conversational AI agents understand customer intent in natural language, query CRM, ERP and legacy systems during the conversation and resolve requests end to end, by text or by voice.

GBPA designs and deploys these journeys on the Cognigy platform, now part of NICE, takes the agents to the channels customers already use with Infobip (WhatsApp, SMS, RCS, email and voice) and gives service a natural voice with ElevenLabs. We define what the agent resolves on its own, when it asks for confirmation and when it hands the conversation to a person.

One of our Cognigy projects, for a large sanitation company, put a WhatsApp chatbot live in six hours during the floods in southern Brazil and was awarded at the Cognigy Awards 2025, in the Most Supportive Use Case category.

Front flow

CX

voztextointençãoagentehumanoCRM

What we deliver

Conversational agents that resolve

Intent understanding, back-office system access and first-contact resolution.

High-fidelity synthetic voice

IVR and voice service natural enough not to push the customer straight to a human.

Handoff with full context

When the human steps in, they step in knowing everything the conversation already covered.

Technologies we implement

METHOD

How we run the project

  1. 01

    Journey analysis

    We analyse contact reasons, volumes per channel and what currently needs an agent to be resolved.

  2. 02

    Agent design

    We define intents, system integrations, autonomy limits and the rules for handover to a human.

  3. 03

    Channel deployment

    We build the agent, connect WhatsApp, voice and digital channels and test with real conversations before opening it to everyone.

  4. 04

    Measurement and evolution

    We track first-contact resolution, handover and satisfaction, and expand what the agent resolves.

FAQ

Frequently asked questions

What is the difference between a chatbot and an AI agent?

A traditional chatbot follows menus and canned answers. An AI agent understands natural language, keeps the context of the conversation, queries systems and takes actions, such as issuing a duplicate bill or opening a ticket, without the customer navigating options.

Can AI handle customer service on WhatsApp?

Yes. It is one of the most common uses. With Infobip, the AI agent serves customers on WhatsApp and other channels with the same history, and GBPA integrates the conversation with the company's systems so the request is resolved right there.

Can AI answer the phone?

Yes. The same agent logic serves voice, in conversational IVRs. With ElevenLabs, speech has natural intonation and rhythm, which reduces requests for a human just because the machine sounds odd.

What happens when the AI cannot resolve something?

The conversation goes to an agent with the full context: who the customer is, what they asked for and what has already been tried. The autonomy limits are set during the project, before the agent goes live.

Why Cognigy?

Cognigy is an AI agent platform for voice and text customer service, built for large contact centres and integrated with company systems. It is now part of NICE.

Shall we look at your operation?

The maturity assessment reads your processes and shows where automation and AI pay back first.