← Cases
AustraliaEnergy and utilities2020 – ongoing

AGL Energy

AGL Assistant virtual agent for customer chats

AGL Energy's virtual assistant handled about 36% of customer chats without a human agent

WorkingVendor case studyVendor ↗

Problem

AGL's first customer chatbot, released in 2017, could only help customers update account details. Many routine questions and transactions still went to frontline agents.

Approach

In early 2020 AGL replaced the old bot with a virtual assistant built on IBM watsonx Assistant, later named the AGL Assistant. The Australian integrator ISW helped connect it to AGL's ERP software through Watson Actions so it could carry out transactions. AGL added features over the following years, such as adding concession cards digitally, and was exploring generative AI.

Results

36%share of chats contained in the virtual assistant, per IBMVendor case study[1]

Takeaway

Linking the assistant to back-end systems let it finish tasks rather than only answer questions, which is what kept chats away from agents.

Categories

Business function
Customer service
Innovation type
Process innovation
Evidence
Vendor case study

Sources

  1. AGL Energy
    IBM10 Jan 2025Vendor publication

Source summaries

[1]IBM10 Jan 2025

The page describes how AGL moved from a basic account-update bot to an IBM watsonx Assistant virtual assistant. About 125,000 customers start a chat each month. Roughly 36% to 38% of these chats are served without a frontline agent.

Full details are in the original sources.

Added 4 Nov 2026 · Updated 4 Nov 2026
Open case page ↗