Generative AI is reshaping how companies listen, learn, and engage with customers. Machines can now understand natural conversations, speak in a local accent, and provide insights in real time.
As Chris Connolly, director, solutions engineering, Apac, at Twilio, explains, this shift is moving quickly from concept to daily reality for retailers, banks, and other businesses interacting by phone or online.
“Every voice, video and text conversation is now being transcribed and used to build data,” he says. “That allows us to mine value that was previously invisible – signals, topics, even intent – by building trust among customers and creating moments that resonate with audiences.”
Just a few years ago, businesses leaned on generic third-party data lists, competing for attention with the same recycled profiles and untargeted ads. Customers were bombarded and quick to switch off. Conversational AI changes that equation. Every interaction becomes a unique data point – one that can be personalised at scale, down to an audience of one.
Connolly gives a practical example: “Now I know that Steve not only lives in a specific postcode and likes to navigate these pages, but the last time he spoke to someone in my organisation, he mentioned that he’s going on a trip in three months’ time. Great. That’s a new piece of data that we did not have before that was mined out of a conversation. So we can start to build these golden profiles in customer data platforms that can really engage with our consumers.”
From clunky chatbots to conversational AI
Businesses have been quick to recognise the opportunity. In Australia, nearly every enterprise has already invested in AI or has a project underway. Connolly notes the business case is clear: Operational benefits and reduced costs. The next step is ensuring consumers feel the benefit, too.
“Australian consumers are still a little bit wary of AI and how their data is being used,” he says. “They’re not all convinced yet, and rightly so.”
Twilio’s annual State of Customer Engagement Report shows the divide. While 97 per cent of businesses see clear benefits from AI, only 46 per cent of Australian consumers agree.
Part of the challenge is overcoming the poor reputation of earlier chatbots. “Chatbots have been around a long time, and they’ve gone through a series of iterations, but unfortunately, we trained the end consumer to ignore them because they weren’t able to service most of their needs.”
Today’s conversational AI is different. It can understand accents, engage in nuanced exchanges, and handle real customer issues. These systems are not just answering questions; they are resolving problems and producing data that feeds back into the customer experience.
Connolly says the difference is striking in areas like collections. “Automated AI conversations work exceptionally well with debt collection because consumers are more likely to agree to a promise of payment with a bot than a human, because of the embarrassment factor. It doesn’t feel as confronting to talk about debt with a bot as it does with a human, where you might have a few more guilt angles thrown at you.”
Startups and enterprises in action
Australian companies are experimenting with a wide range of use cases.
Startup Driva, a broker helping younger consumers finance their first car purchase, built its original model to be fully digital. It later layered in AI using Twilio capabilities to answer basic questions. The company has now found the right balance between automation and human interaction – using AI where it adds value, then passing more complex cases to staff.
An energy retailer is trialling voice AI in its contact centre to reduce friction. Callers speak with a natural-sounding Australian-accented bot that resolves many issues outright. When calls are escalated, the bot provides the human agent with a summary of the conversation, so they can begin at the right point.
“This company has seen true benefits immediately,” Connolly says. “Getting their customers to the right place more quickly … might lead to saving only a few seconds here and there, but it adds up quickly, especially when you are receiving thousands of calls every week.”
Retail is also seeing benefits. A pizza chain fed hundreds of thousands of call recordings into Twilio’s AI platform, segmented them by location, and quickly saw which outlets triggered the most complaints.
“You can dig into quality issues, like if a driver did not arrive, or was late, or the pizza was delivered cold,” Connolly says. “We discovered there were certain stores that didn’t answer calls; that is money left on the table for that franchise, and an unhappy consumer who may not call again.”
Trust and transparency
With every conversation turning into a data source, transparency has become critical. Companies must be clear about how data is stored and used.
“This is starting to hit the forefront of consumers’ minds here,” Connolly says. “Most people have used ChatGPT or something similar by now. So there is an awareness that this technology exists. The number one challenge is getting consumers to trust that AI is going to deliver value and that there is going to be a value exchange.”
Disclosure is one area where companies take different approaches. Connolly points out that one customer that presented its bot as human saw five to 10 times the engagement rate of another that was upfront about using AI.
However, he argues, transparency remains the right course. “Transparency is key. You should always inform someone when they’re interacting with the chatbot, so that they can make informed choices and give informed consent.”
He sees consumer behaviour shifting. As people realise bots can actually solve their needs, they are more open to engaging. Twilio advises its customers to make disclosure part of the user experience.
“We absolutely advocate for transparency in where automation has been used and where it’s not,” Connolly says. “We recommend adopting AI nutrition facts, which are very much like those used on food packaging. If you pick up a can of soda, you will see its sugar content, sodium, and other ingredients listed. We think that the same type of package labelling should be used for software products.”
Brands can create their own AI nutrition labels on Twilio’s platform, detailing policies such as whether humans are in the loop, how long data is retained, and whether a model is predictive or generative.
An even more practical disclosure method is to give customers a choice. Connolly suggests informing callers they can speak to a human, but it may mean a longer wait, or they can try the AI assistant immediately.
“You get very good adoption,” he says. “You also find that many problems are resolved without human intervention, leaving staff free to deal with more complex issues, or you very quickly work out an issue that is urgent, which needs to be handled by a human.”
From novelty to necessity
In Australia and elsewhere, businesses are embedding conversational AI into the heart of their customer engagement. The technology is not replacing people, but it is changing how work gets done, making every conversation a new opportunity for insight.
For Connolly, the direction is clear: The winners will be those who combine the efficiency of automation with the trust of transparency. “Modern conversational AI isn’t just about answering questions,” he says. “It’s about creating a relationship where every interaction adds value.”
