AI promises faster, cheaper customer service. But are chatbots ready for prime time? This week’s Let’s Talk asks experts if businesses should trust AI with customers.
Customer service stands at a technological crossroads. AI-powered chatbots can now understand natural language, resolve common issues, and operate around the clock without breaks.
For businesses, the appeal is obvious: reduced costs, faster response times, and scalability that human teams simply can’t match. Yet customer frustration with automated systems remains high, and the risk of damaging valuable relationships looms large.
In this week’s edition of Let’s Talk, we gathered insights from experts across technology, business strategy, and customer experience to explore a question every modern business must answer: Should you let AI talk to your customers?
Let’s Talk!
Ginger Kidd, VP Marketing and Communications APAC, Sinch
“AI is a valuable tool for enhancing the digital customer experience, but only if it’s used effectively and with a customer-centric approach.
“Despite AI becoming more broadly used and accepted, Australians are cautious. Sinch’s State of Customer Communications Report found that two-thirds (66%) do not trust the accuracy of AI, and only 27% are willing to use an AI solution for customer support, even when it is trained on the business’s own documentation.
“In a nutshell: yes, brands should let AI talk to their customers, but use it strategically to streamline straightforward interactions while providing easy access to human support. AI works best when it supports human interactions rather than replaces them.
“Before introducing AI, consider whether your audience is receptive or if it could risk undermining trust in the brand. Once implemented, success will come from being transparent about AI use, maintaining strict privacy standards, and offering customers choice. Brands that find the right balance and put customer trust first are more likely to build loyalty and stand out in the market.”
Apoorv Iyer, Head of GenAI/AI Practice, HCLTech
“Allowing AI to interact directly with clients brings significant advantages but it also comes with its share of risks and limitations.
“There’s no denying the value AI-powered agents offer instant responses and 24/7 availability drastically cut down wait times and help businesses support customers around the clock. In fact, Gartner has reported a 30% drop in operational costs and a 25% boost in customer satisfaction for companies using AI, thanks to faster resolutions and personalised experiences. McKinsey also found that AI chatbots can manage up to 70% of customer interactions without human intervention, significantly improving scalability.
“However, we must also remain cognizant of the limitations and some risks associated with this strategy. AI, as advanced as it is, lacks emotional intelligence, which is crucial for sensitive or complex issues and interactions. Many customers prefer human agents for nuanced issues. Implementing AI also requires investment and privacy safeguards. We at HCLTech believe that the best approach is a hybrid model wherein AI manages routine responses efficiently, while humans handle more complex or sensitive cases which require human intervention.
“A great example of this in action is a Global 2000 financial services company that uses AI-powered virtual assistants to answer thousands of client queries daily, reducing resolution times significantly while allowing human agents to focus on complex cases. This blend of AI efficiency and human empathy ensures seamless service quality.”
Nicole Birbas, Director of Customer Success, Klaviyo
“AI has progressed to the point where businesses should allow it to talk to customers — especially retail brands who are dealing with hundreds if not thousands of transactions each day. For these brands, customer service is the difference between a make-or-break relationship with customers.
“But that doesn’t mean AI should replace human connection — instead it’s about scaling it. Our research shows that 63% of consumers expect AI assistants to be a normal part of online shopping by 2026, and increasingly want AI to help them find the best deals and get the most value. Brands embracing this shift are setting a new standard for service, speed, and loyalty.
“When used well, AI eliminates friction by answering questions instantly, recommending products in real time, and freeing humans to focus on more complex problem solving. The most successful brands with a competitive edge won’t be the ones automating the most, but the ones blending empathy with intelligence.
“At Klaviyo, we’re seeing this shift first hand through tools like Customer Agent, which show how AI can extend a brand’s capacity for personalised service — proving connection, not convenience, is the true currency of customer trust.”
Christopher Connolly, Solutions Engineering Lead, Communications, APJ, Twilio
“Most people, knowingly or not, are already dealing with AI in their day-to-day life – whether ordering some take away, paying a bill or listening to music.
“As AI continues to evolve, it drives more efficiency gains and streamlined customer service for businesses. A generative AI chatbot can write a perfect reply to a refund request, but agentic AI can now also proactively process the refund and follow up with the customer a week later to close the loop. But getting AI right needs careful consideration and the right tech tools. Twilio’s recent Decoding Digital Patience research shows almost half Australian consumers get frustrated with automated voice and AI chatbots, compared to around 10% with phone calls to human agents.
“The key to ensuring the best AI experience for your customers comes down to two things: transparency, and autonomy. Brands need to be transparent with their customers about when they are engaging with AI, or their data is being used to train it. While AI interactions are now much more human-like – with conversations flowing quickly – customers always need an off-ramp. Fortunately, advanced Voice AI tools can now detect interruptions, letting users cut in for quicker answers, and can also sense when to transfer complex issues to a live agent.”
Richard Valente, Executive Vice President, Customer Experience Strategy at TP in Australia
“AI is reshaping how we interact with customers – but I don’t believe it should ever fully replace the human touch.
“At Teleperformance (TP), we see the best results when AI and people work together. AI is brilliant for handling routine tasks and improving speed, accuracy and availability. But when it comes to empathy, tone and understanding cultural nuance, nothing beats human connection.
“I always tell clients: let AI start the conversation, but let people finish it. That’s how you create genuine engagement. When a customer feels heard, they’re far more likely to stay loyal to your brand.
“AI should support, not substitute, emotional intelligence. The goal isn’t to automate relationships, but to empower human agents to focus on complex, emotionally charged interactions – the moments that matter most.
“We also encourage businesses to test their own systems regularly. Experience what your customers experience. If it feels impersonal or robotic, it probably is.
“In the end, technology can make customer service faster and smarter, but it’s people who make it meaningful and build brand loyalty.”
Jonathan Tanner, Senior Director, Industry Principal Financial Services & Insurance APAC at Pegasystems
“AI in customer service has seen explosive growth in recent years and for good reason: with the rise of AI chatbots and assistants excelling at handling routine inquiries such as basic Q&A or order tracking, it frees human time to focus on more complex and high-value interactions.
“So the real question businesses should be grappling with isn’t simply whether they should let AI talk to customers, but when and how AI should be involved.
“For repetitive and low-stake tasks, customers often prefer the efficiency of automation over waiting for a human agent. Meanwhile, for complex and high-stake situations that may impact customer trust, human judgement and empathy are crucial. The key is knowing where to draw that line.
“So yes, let AI talk to your customers, but establish clear guidelines and maintain human oversight for complex issues so you don’t risk trading customer loyalty for efficiency. The goal isn’t automation for automation’s sake, but delivering better, more personalised experiences at every touchpoint.”
Emma Seymour, Chief Financial Officer, Deputy
“Whether or not you should let AI talk to your customers really depends on the kind of experience you want to create. At Deputy, we don’t see AI replacing human connection — we see it amplifying it. The best customer experiences happen when technology supports people to do what they do best.
“AI can analyse patterns, anticipate needs, and handle repetitive tasks with incredible speed. But empathy, emotional intelligence, and the ability to read the moment — those are deeply human skills, and they’re still what customers value most.
“In a recent survey we just conducted 68% of shift workers using AI said it helps them deliver a better customer experience, and 71% of managers said it makes interactions more consistent and reliable. When AI takes care of repetitive, time consuming tasks, it gives teams the bandwidth to focus on customers — listening, problem-solving, and building trust.
“And while AI is becoming more capable every day, 94% of workers still believe empathy and care can’t be replaced by technology. That’s why the smartest businesses aren’t choosing between AI or people — they’re combining both. AI can be part of the conversation, but people should always lead it. That’s where real connection — and long-term loyalty — is built.”
John Brown, Senior General Manager, Strategy, AI and Emerging Vendors, Ingram Micro
“Despite the profound impact AI can have, bad AI prompts used by an SME – like misleading ads or rogue chatbots – may spark breaches of law, defamation against a third party, reputational damage or even lawsuits which may trigger financially devastating consequences.
“Strong AI governance is now as critical as cyber security. Input controls, human oversight and isolating critical systems are standard controls that SMEs should consider putting in place and form part of their offering in guiding end customers through AI adoption.
“Responsible AI use isn’t just about compliance; it’s brand insurance. Businesses that build their AI strategy on security, transparency and human judgement will be the ones that earn lasting trust.”
Bede Hackney, Head of Zoom ANZ
“A recent study commissioned by Zoom into Australians aged 18 to 24 who have had early exposure to AI and are currently active users, revealed this cohort is seeking seamless integration between AI and human expertise. While this group is dubbed ‘AI natives’, we found that while 59% of AI natives said businesses should offer AI agents for faster resolutions, four in five (80%) still wanted the option of escalating to a human agent in their customer interactions.
“Among those who consider this important, the top two reasons cited are that AI responses are often too generic and unhelpful (63%), and that they feel more confident that a human agent will correctly understand their issue (59%). In fact, getting generic copy and paste answers were noted by Australia’s AI natives as the top factor negatively impacting brand loyalty (53%).
“While the benefits of scale and cost savings dictate that most organisations will have AI engaging their customers, companies must prioritise delivering high-quality, accurate engagements. And finally, rather than looking to completely displace their human agents, instead redeploy those valuable human agents to the points in the customer journey where empathy and human connection can deliver the greatest impact.”
Michael Hou, Solutions Architect, Rackspace Technology
“The promise of AI in customer service is appealing. It can deliver rapid responses, reduce costs, and provide a level of availability most small businesses could never staff on their own. But deciding whether to let AI speak directly to customers depends on what those customers are asking of you.
“Customer expectations have moved beyond static chatbots that simply repeat pre-programmed answers. They now look for adaptive, agentic experiences that respond intelligently to intent and context. For routine queries such as booking confirmations, product information, or account balances, AI can perform this role effectively. It provides instant service while freeing staff to focus on more valuable work.
“The challenge emerges when queries are complex, emotional, or require tailored advice. In these moments, authenticity matters. With AI-generated communication increasingly indistinguishable from human interaction, businesses risk damaging trust if customers feel misled or poorly handled.
“The most effective model is layered. Allow AI to act as the assistant that filters, guides, and resolves straightforward tasks, while ensuring more advanced queries are seamlessly transferred to a human. By combining AI efficiency with human judgement and reassurance, businesses can meet rising expectations without sacrificing trust.”
Rich Atkinson, Executive Director of Technology, Airteam
“AI Can Talk to Your Customers, But Only If Humans Are Standing By.
“AI should talk to your customers, but the key is matching the right enquiry to the right solution, and always providing a clear path to human support.
“Recent research from McGill University, analysing over 500,000 customer service interactions, reveals the critical gap. While 71% of customers prefer human agents over chatbots, and 60% report chatbots often fail to understand their issue, the problem isn’t AI itself, it’s how businesses implement it.
“The research found customers actively avoid chatbots for detailed or sensitive enquiries, particularly shipping and payments, preferring humans when complexity increases.
“The solution is strategic deployment. Before routing a customer, identify enquiry complexity and sensitivity. Simple, routine questions? AI excels with 24/7 availability. Complex, nuanced, or sensitive issues? Route directly to humans.
“Most critically, code seamless escalation paths into your systems from day one. When chatbots fail, customers need immediate human access, not endless loops.
“Think hybrid, not replacement. Deploy AI for routine tasks, but respect the majority customer preference for humans when it matters most.”
Jarrod Kinchington, Vice president Australia and New Zealand, Smartsheet
“In today’s rapidly evolving AI-powered landscape, the question isn’t simply whether to let AI interact with your customers, it’s how you enable and govern that interaction. At Smartsheet, we believe AI works best as a collaborative teammate, enhancing decisions, streamlining tasks, and offering smart suggestions, all while keeping a human in control.
“Generative AI tools are increasingly capable of producing summaries, visualisations, and recommendations. But the true value lies in explainability; AI that can show its work. For example, when asked to build a report or chart, responsible AI systems should not only provide the output, but explain the reasoning behind it: what data was referenced, how it was interpreted, and what criteria drove the result. This level of transparency builds trust, particularly in customer-facing scenarios.
“This “co-pilot” model is essential, especially in regions like Australia and New Zealand where businesses are embracing AI but remain cautious about control, privacy, and ethical use. AI should not be a black box speaking on your behalf; it should be a transparent, auditable, and accountable partner.
“Ultimately, the future of AI in customer engagement isn’t about replacing people, it’s about empowering them.”
Sharon Nouh, Founder and CEO at ProSpend
“Yes, you should let AI talk to your customers, but on your terms. Treat a customer-facing bot like a new team member: trained, supervised and tightly scoped. Under Australia’s Privacy Act, anything customers type into a chatbot – and any output containing personal information – triggers your privacy obligations, so you need due diligence, human oversight, and clear disclosure that it is a bot. The privacy regulator also advises against putting personal or sensitive information into public generative AI tools at all.
“And, we’ve just seen why. A NSW contractor uploaded thousands of flood victims’ details into ChatGPT – an avoidable breach that underscores how easily well-meaning staff can mishandle data if the rules aren’t crystal clear. Build policy first and deploy the bot second.
“Security and consumer law cut the same way; a bot must never be a single point of failure. LLMs can be manipulated via prompt-injection and can take excessive, unintended actions if you grant them too much autonomy, so keep them on a short leash and have humans test and validate outputs before anything is sent or actioned.
“For higher-risk moments – identity, refunds, complaints, anything that could create a legal commitment – escalate to a human and give customers a clear path to a person.
“Digital tools that mislead, even by hallucination, put you on the hook under the Australian Consumer Law, and regulators are pushing for stronger protections across digital platforms.”
Jonathan Reeve, Regional Director, ANZ at Eagle Eye
“Customer personalisation is the new must-have element of any sophisticated operation. While customer satisfaction has always been a goal of the retail and service sectors, the level of expectation among customers has escalated in line with the efficiencies that modern digital technology has brought. For example, Eagle Eye already has a powerful AI-driven personalisation engine and other predictive systems, which thrive on ingesting and processing data intelligently.
“Large language models (LLMs) have evolved rapidly, now performing beyond pattern-based text recognition. Just think about how much share of research LLMs have already taken from traditional search engines. In addition to being able to ask questions, AI agent helpers can make decisions, compare prices and steer people to where to shop. This stands to change how retailers reach customers.
“As with all technologies, AI has upsides and downsides. Retailers will succeed if they understand how agents evaluate and present options to consumers. While AI increases engagement and personalisation at scale, brands must use it strategically, remain transparent, and preserve the human element to maintain consumer trust.”
Billy Loizou, Area Vice President and General Manager, APAC at Amperity
“In Australia, the use of AI is growing for marketing, eCommerce and digital leaders across all industries, but many are still at an early stage.
“To succeed with AI, retailers must create solid customer data foundations first. This means deploying platforms that collate various information sources in real-time and present unified 360-degree customer profiles.
“AI runs on customer data. Customer data platforms (CDPs) make that data usable at enterprise scale. Retailers with a CDP are twice as likely to feel confident about understanding and acting on customer behaviour, and three times more likely to deploy AI in customer-facing applications.
“These platforms pull information together and make sense of it for AI applications. They tackle major customer data challenges, including the problems created by duplicate profiles. Duplicate profiles create real problems when retailers engage with consumers across various touchpoints. Duplication happens when people use different identifiers across separate channels, abbreviated names, alternative email addresses, or different contact details.
“Unifying customer data requires identity resolution. Without it, personalisation, consent management and omnichannel activation become much harder. It also improves AI application accuracy. When AI personalises interactions, effectiveness improves dramatically when working from complete profiles. It enables cross-channel engagement and prevents customers from receiving conflicting communications.
“A better data foundation will help brands move faster, make better decisions and improve their growth potential. Clean consolidated and organised data will enable AI to take on more work, generate better insights and help brands connect more meaningfully with their customers.”
Karim Rayes, Chief Product Officer at Nexxen
“The power of AI is fundamentally elevating every stage of the advertising life cycle. On the engineering side, AI assistants typically have multiple agents tied to them, leveraging large language models (LLMs) to solve specific problems. I expect that over the next few years, the AI-automation initiative will continue and get to a place where most of the workflows can be fully automated.
“At Nexxen, we are among the first in market with a completely integrated solution that works seamlessly with our DSP. As a global advertising technology platform with deep expertise in data and advanced TV, we are deepening the impact of customer data insights and discovery for business with our latest advancement. Our nexAI Discovery assistant enables clients to quickly turn complex consumer data into clear, actionable audience profiles and campaign planning for seamless activation.
“The Nexxen Data Platform has always been powered by advanced machine learning to help our clients navigate the fragmented media landscape. It’s about removing friction from the entire workflow, enabling advertisers and agencies to move from insights to activation faster, smarter, and with greater confidence.
“Ultimately, companies that treat AI as core infrastructure, not just another shiny add-on feature, will lead the way.”
Dan Pearce, General Councel, Holding Redlich
“Before deciding whether to introduce an AI chatbot to your customers, carefully consider which tool you will use and who provides it. Does the supplier have experience in your market and offer support during your business hours? Have you tested how the chatbot operates and what rules it follows when generating a response? It is vital to understand what data the tool has been trained on and where it has come from, as public generative AI tools often rely on vast datasets that may be biased, unreliable or include copyrighted material.
“You should also consider what information you or your customers will share and whether that data might be used by the supplier to further develop the tool. For sensitive or highly important decisions, keep a human in the loop and ensure transparency, especially if you need to justify an outcome later.
“Make sure users know they are interacting with AI, can challenge its outputs, and that you have clear records of accountability and policies in place. Even something as simple as a chatbot can cause serious issues if not managed carefully, so working through this checklist helps reduce the risk and protect both your customers and your business.”
Rahul Tabeck, Regional Sales Director of ANZ, SolarWinds
“AI customer engagement isn’t exactly new. What’s next is AI agents that plan and act, not just reply. The smart move is to draft a roadmap now. Treat it like any other critical system: instrument first, scale second. Move in phases with safe experimentation, full stack observability, predictive operations and compliance-driven controls. Start with assistive and low-risk journeys such as password resets, status updates and simple FAQs. Tie the agent to a single source of truth, define SLOs, confidence thresholds and a kill switch.
“Observability must be part of the agent’s scaffolding, not just the infrastructure layer. Every action should be traceable, explainable and reversible from day one. A mistuned agent can throttle the customer experience, misroute enquiries or take unintended actions in seconds.
“Govern what you can see and only trust what you can audit. If you cannot explain an action path, it should not be in production. Do this well and AI frees up your teams for higher-value work while customers get faster, more consistent service.”
Elise Balsillie, Head of Thryv Australia and New Zealand
“Artificial intelligence used by small business has become the quiet force behind many customer interactions – from chat support to appointment scheduling. However, the question for small businesses isn’t whether AI can talk to your customers, it’s whether it should.
“The short answer is yes, however, only with structure, strategy and empathy. AI can help your business respond faster, manage volume and maintain consistency across every channel. It can also answer FAQs at midnight, confirm bookings instantly or flag when a customer’s message needs a personal touch. The key is balance – allowing automation to handle routine tasks, while keeping your human voice where it matters most.
“The most effective AI is there to reinforce relationships, not replace them. When AI tools are trained on your business’s tone, values and service standards, they become an extension of your business, not a barrier. Customers should feel heard, not handled.
“Before switching AI on, I recommend that you map the journey: where does automation add value and where must people remain present? When used thoughtfully and strategically, AI can build reliability and consistency, freeing you to focus on the deeper work of connection, service and growth.”
Paul North, General Manager for Asia Pacific, UserTesting
“AI is no longer optional, it’s already a key part of the customer experience. AI chatbots, for example, are increasingly replacing the earlier phases of the shopping journey, helping with brainstorming, research and cost comparison. In fact, recent research we did around Black Friday spending revealed that nearly half of Aussies surveyed now use AI to shop.
“But building AI that customers actually find useful takes time. In the race to integrate AI into the user experience, many businesses skip the crucial step of validating the customer experience. When AI feels robotic, unhelpful, or unclear in its reasoning, it doesn’t just miss the mark, but risks disengaging customers or damaging trust.
“To avoid this, it is important to validate AI-driven experiences with customers. AI should complement employees and customers during the sales process, enhancing the interaction rather than replacing it. Businesses that get this right will set the standard for AI-driven customer experience.”
Julian Fayad, Founder and CEO of LoanOptions.ai
“Businesses should absolutely embrace AI to talk to their customers, but we have to get the purpose and conditions right. It’s about taking a hybrid approach.
“Rather than replacing human agents completely, AI is best utilised as a tool to support them, particularly with high-volume, low-complexity interactions. AI can remove the drudgery of repetitive tasks, completing them with superior speed and accuracy. AI can also speed up the lead qualification process, or follow up supporting documents from a client, and serves to eliminate missed opportunities and ensure your human sales team only spends time on the highest value-add interactions with high-quality leads. There are other use cases including retention at scale. Having touch points strategically within the life cycle can serve to identify the highest intent opportunities.
“For customer-facing AI to work, stringent evaluation and monitoring are non-negotiable. We need processes like keyword and sentiment analysis and repetitive query recognition to trigger a seamless handover to your human team the instant an interaction requires any more advanced skills.
“While AI delivers predictable answers and efficiency, it can’t offer the genuine trust, emotional support, and human judgment necessary to maintain loyalty. That strategic partnership between AI efficiency and human care is simply the best practice for modern service.”
Peter Curran, Founder & Business Development Manager, Digital Surfer
“There’s a lot to unpack around using AI to talk to your customers. The first thing is keeping in mind around 65% of a business’s revenue often comes from repeat customers, and 8 out of 10 Australians say they will go somewhere else if they have a poor customer experience. So, you want to make sure if you are using AI to talk to customers, it’s been set up to answer as many queries as possible or make it easy for them to escalate to a human if their query can’t be resolved. There’s nothing worse than wasting time going round in circles with AI chatbots.
“The upside is that using AI, like implementing chatbots on your website, can improve response times and make sure people can get an answer 24/7. 90% of businesses report faster complaint resolution when they use chatbots, and a decrease in call volume too, giving your team time to focus on other areas.
“You can also use AI to personalise an experience, which can improve retention rates even further. You can use AI to recognise past behaviours or purchases of people on your site to make recommendations, or predict what they might be needing.
“If you’re smart with it, AI can really help with the customer experience you provide, but you have to be investing the time and resources to make sure it is an amazing experience to not miss out.”
Greg Wilkes, CEO of Develop Coaching
“AI is creeping into every corner of business, from your estimating software to your email inbox. But should you really let it speak directly to your customers? The short answer: yes, but with guardrails.
“AI can handle routine stuff brilliantly. Think FAQs, scheduling, follow-ups, or simple quote requests. It’s fast, never sleeps, and doesn’t get tetchy when a client changes their mind. For many firms, that means faster response times and better first impressions.
“But let’s be honest: AI still lacks emotional intelligence. It can’t read between the lines when a client’s stressed about their kitchen refit or nervous about delays on site. If you hand over all your client communication, you’ll lose the human touch that wins trust and repeat business.
“The smart move is hybrid. Use AI to do the heavy lifting, then step in when the conversation turns emotional, complex, or high-value.
“So, don’t fear AI. Train it, test it, and make it part of your client care system not your replacement for it. The businesses that balance efficiency with empathy will own the next decade.”
Leanne Shelton, CEO, HumanEdge AI Training
“If you want your customers to have a positive experience with AI, the tools must be used with thought and transparency.
“Trained well, AI Chatbots and Phone Agents can save your team from answering the same questions over and over. They can also be taught your FAQs, connect with your CRM, and capture lead details mid-conversation. But while they can talk to your customers, let’s not pretend they can replace your sales or service teams.
“The best Chatbots and Phone Agents seamlessly pass the conversation to a human when things get complex – because no one wants to go round in circles with a bot that won’t hand over the call.
“The real magic happens when AI handles the repetitive, and humans handle the relationships. That’s how you build trust – and keep it.”
Jayden Rozsahegyi, Digital Solutions Lead, Hatched
“Should AI talk to your customers? It all depends on what the query is. AI responding to customers without a QA process – a system to check the responses before the customer sees them – I’d argue, is 100% a no.
“An AI agent trained on a specific set of brand-safe guidelines and knowledge bases might be the key to having a chatbot that can answer quick queries. I’d start with what can save your brand time. Is it something as simple as how do I update my password on their system, or is it more in-depth, like advice on specific guidelines?
“It’s important that we all take a cautious approach to AI implementation. However, by creating a workflow for simple tasks and automating basic FAQs, you can save a notable amount of time.
“That said, I would advise business owners against implementing a platform with an AI that would respond to random queries, as that can come with a lot of nuances.”
Peter Traianou, Co-Founder of Waave
“The short answer is ‘yes’, but it’s a little more complex.
“Agentic AI has opened the door to autonomous experiences that will redefine commerce. These changes will infiltrate not only customer facing services, from marketing, to retail journeys and digital payments, but also the back end, such as helping a banking risk analyst surface fraud anomalies across merchant customers, assisting a banking onboarding specialist with application reviews, or enabling a payment operations manager to generate reports instantly.
“Already in retail, advancements by players like OpenAI are cutting out the middleman with features like ‘Instant Checkout’ where customers can complete purchases directly within the ChatGPT. Businesses therefore cannot afford to stall on integrating AI capabilities through branded agentic experiences.
“However, security is vital in maintaining customer trust. This involves investing in technology partnerships and components that create a secure environment where AI agents can act on behalf of merchants or consumers, turning natural language commands into precise API calls. For example, we integrate an AI co-pilot directly into our merchant control centre. By orchestrating complex, multi-step payment and commerce workflows, it not only opens the door to autonomous commerce experiences; it also follows the same RBAC rules as human users, ensuring least-privilege access. All activity is logged in immutable audit trails, maintaining transparency, compliance, and non-repudiation. Businesses need to go about AI the right way, and leverage that credibility to instill customer trust so they are comfortable using it.”
Victor Horlenko, Head of AI Innovations at Devart
“Absolutely, but don’t hurry and do everything right.
“In reality, AI can significantly enhance your customer experience, offering 24/7 quick availability, instant and accurate responses, and reliable assistance. There is no need to wait hours to get help, now it’s a matter of seconds.
“But always keep in mind that any AI system relies not only on AI, but on the data that supports it. Without precise knowledge, or if the chatbot was trained on outdated information, your users will end up frustrated with zero real assistance. Before any AI integration or tool deployment, do not be afraid of investing serious time in creating and clearing your knowledge base, and perform a lot of tests internally to be sure everything is covered.
“Visualize this as a new member onboarding- the better training and better results. This is the same case – if everything is done correctly, AI becomes one of the most reliable agents: fast and scalable.
“To cut a long story short, yes, AI can talk to your customers, make sure it speaks the language of your product and company.”
Kathryn Giudes, Founder and Managing Director at ORCA Opti
“You can let AI talk to your customers, but only if you’re ready to take full responsibility for what it says.
“Research shows that AI has proven its worth in customer service, cutting response times by up to 74 percent and slashing costs from around $13.50 per assisted contact to just $1.84. Companies like Klarna, Virgin Money and NatWest have demonstrated that when implemented strategically with strong guardrails and human oversight, AI can boost satisfaction and deliver multimillion-dollar returns.
“The risks of AI customer service are not theoretical. They're documented, expensive and occasionally, catastrophic. Air Canada learned in 2024 that you can’t blame the bot when things go wrong. Businesses are legally liable for AI’s mistakes. The January 2024 DPD incident showed the world how fast AI can turn from a helper to a hindrance. After a system update, a customer manipulated the chatbot into swearing and calling DPD “the worst delivery firm in the world,” urging users to avoid the company. The viral screenshots, which were viewed more than a million times, forced DPD to shut the bot down.
“The risk lies in deploying AI without the adequate safeguards in place. It’s considered best practice to use AI to handle routine, high-volume queries while humans manage the complex and emotional ones. Build in transparent escalation paths, continuous monitoring and specialised security infrastructure to defend against manipulation.
“Letting customers talk to AI can be transformative, but only if you treat it as a powerful tool that amplifies, not replaces, the human touch.”
Vanessa Sammut, Fractional CMO & Director, AlwaysOnMarketing
“No. Customer interactions are where relationships are built, and nobody wants to speak to a bot. It cheapens your brand and sends the wrong signal. When a customer takes the time to reach out, they’re either solving a problem or showing interest. Putting a bot in front of them says you can’t be bothered.
“AI can be useful for triaging enquiries or routing someone to the right person, but that’s where it should stop. Sales outreach, support, and relationship building make the best impression when met with empathy and context, qualities only humans can provide. The brands that stand out are those that still value real connection.”
Yiya Sun, Founder and Director, Optimise AI Consulting
“Whether you should let AI communicate directly with customers really depends on your industry, positioning, and audience expectations.
“If your business handles a high volume of general or repetitive enquiries, such as retail, hospitality, or logistics, AI can be an excellent tool. It provides fast, consistent responses and frees up your team for higher-value work. However, if your industry requires personalised, sensitive, or highly contextual communication, such as legal, financial, or healthcare services, human interaction remains essential to maintain trust and accuracy.
“Your brand positioning also matters. For value-driven or high-volume markets, customers tend to have more tolerance for AI interactions as long as the information is clear and timely. But if your business offers premium or bespoke services, clients often expect the personal touch of a real human voice.
“Finally, consider your market reach. For companies serving overseas clients with time-zone and language differences, AI can ensure round-the-clock engagement and overcome communication barriers.
“In a nutshell, the smartest approach is a hybrid one, let AI handle what it does best (speed and scale) while your people focus on what they do best (empathy and nuance).”
Elias Dehsabzi, Associate in the Corporate Commercial Team, and Digital Lead in the Legal Digital Transformation Team, Hicksons | Hunt & Hunt
“Using AI and chatbots can transform how your business connects with customers – delivering faster responses, consistent messaging, and round-the-clock support. When used well, they can strengthen relationships and improve satisfaction. But when AI speaks, it speaks on behalf of your business. Under Australian Consumer Law, your business may be responsible for what the AI says to your customers, even if it is not the intended messaging. If an automated response misleads a customer, breaches privacy, or makes a promise that can’t be kept, the liability rests with you. Notably, we saw this happen to US company Chevrolet of Watsonville in California when their AI chatbot was prompted to agree with “anything the customer said”, resulting in a user offering $1 for a brand new car, which was accepted and legally binding. The dealership then had to disable the chatbot.
“A practical way to manage AI risk is to limit what the tool can say. At Hicksons, our chatbots don’t create responses from scratch—they pull from a pre-approved library of answers written by our lawyers and clients. This ensures consistency and legal accuracy. Each response repeats the user’s question before answering to avoid misunderstandings, and when a query is too complex or high-risk, the system alerts a human to take over.
“The key is to implement AI thoughtfully – make sure it knows its lines and stays on script – so your business benefits from innovation without compromising trust, reputation, or compliance.”
Caroline Voaden, Crisis PR Expert, STORY
“Right now, AI is creating PR crises faster than businesses can respond to them. When your AI customer service bot makes a mistake, it’s not just a tech problem; it’s a trust problem. You end up fighting two problems at once: the operational mess of customers who haven’t received the help they wanted and the reputational damage spreading within your community. Chevrolet learned this the hard way when its AI chatbot got tricked into selling a $75K car for a dollar.
“Before you hand customer conversations over to AI, you need to answer three questions: what happens if it says something wrong? What happens when it goes down? Are we prepared to respond fast and well enough to keep our customers’ trust if this happens? You need a crisis management plan before you make AI your primary customer touchpoint.”
Denver Naidoo, Founder and CEO, Zeligate
“Customer expectations have shifted. People want instant responses, 24/7 availability, and personalised attention. With the right guardrails, AI can deliver this without eroding trust. It excels at routine queries, triage, status updates, scheduling, and simple troubleshooting, and it can capture structured feedback at scale. That frees human teams to focus on complex issues and relationship building.
“Success depends on design. Set clear escalation paths to a person, train models on approved knowledge, monitor for accuracy, and disclose AI use where appropriate. Protect privacy, log interactions, and measure outcomes such as resolution time, CSAT, and deflection rate.
“Most customers do not mind whether they speak to a bot or a person if the service is fast, accurate, and courteous. Many firms already use AI to extend coverage and improve consistency. The question is less “should you” and more “can you compete without it” in a market that rewards speed, reliability, and thoughtful human handoffs.”
Richard Taylor, Managing Director, Digital Balance
“The decision to deploy artificial intelligence in customer-facing roles is no longer a futuristic concept but a present-day operational choice. When considering whether to let AI converse with your customers, the answer lies in strategic deployment, not blanket adoption.
“For routine enquiries and 24/7 support, AI is invaluable for streamlining operations and achieving cost optimisation. However, a critical safeguard is limiting its scope. Avoid putting an essentially open Large Language Model (LLM) interface on your public site. The AI must operate within a “walled garden” of verified business knowledge, preventing it from generating irrelevant or incorrect information.
“The second key requirement is that you must implement a clear, effortless path for customers to escalate to a human agent. AI excels at support, but complex issues, emotional interactions, or moments where trust is critical still require the empathy and judgment of a person. AI should be a high-efficiency tool that augments your team, never a gatekeeper that frustrates your most valuable asset, your customer.”
Braith Tomlinson, General Manager – Retention, MYOB
“When customers first engage with your business, whether they’re buying a product, booking a service or signing up online, they expect automation. A guided checkout, product and price comparison tools, confirmation emails, or a scheduling facility are all part of today’s experience. These are table stakes. But when a real person checks in, listens, and adds a personal touch, that’s when loyalty starts to build. The key is knowing when AI adds value, and when human connection matters more.
“AI can make customer experiences faster, smarter and more personalised. It can help your team manage workloads, improve service consistency, and offer insights that guide better decisions. But being customer-centric means using technology to amplify and strengthen your customer relationships, without placing moments where a human touch is required at risk. Many businesses are now focusing on what truly enhances the customer journey. Streamlining the digital experience and creating a seamless customer flow at every touchpoint – both automated and human.
“The most customer-focused businesses use AI to listen and learn, not just to talk. Let AI handle the repetitive, data-driven work, the basic interactions, and let your people lead where empathy, trust and credibility count most. And where a customer may have encountered a stumbling block in their journey, they often want just one thing: to speak to a real person.
“There is absolutely a role for both human and AI customer interactions. It’s our responsibility to ensure we provide both human and tech-focused experiences that deliver the best customer outcomes.”
Discover Let’s Talk Business Topics
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.
