AI was everywhere at NoVacancy Sydney 2026
When I returned to NoVacancy Sydney, the annual exhibition for the Australian hotel and accommodation industry, in September 2026, one thing stood out: AI seemed to be everywhere. Software companies were demonstrating new AI features that promised to save time, improve customer service and help hotels run more efficiently. Some of what I saw was genuinely interesting.
But it got me thinking. Were these companies adding AI because their customers had problems they wanted solved? Or was AI becoming something every software company felt it needed to offer? More importantly, how could businesses already relying on that software tell whether the new AI features were worth using?
It's not just happening in hospitality. AI is appearing in the software businesses use to manage their accounts, communicate with customers, organise work, prepare documents and run day-to-day operations. Often, the new features arrive as part of a software update, sometimes accompanied by new subscription options or updated terms and conditions. You may not have asked for AI. You may not even know what the new feature does. But suddenly it's there, and someone in your business needs to decide what to do about it.
We're increasingly seeing this in conversations with business owners. Recently, someone asked us to look into the security of a new AI feature that had been added to software their business uses every day. Their first question wasn't how to start using it. They wanted to know what had changed, what would happen to their information and whether it was safe. It's a sensible question, and one I think more businesses should be asking.
Why is everyone adding AI to their software?
There are some very good reasons. AI can do things that simply weren't practical with traditional software. It can interpret the meaning of an email, make sense of information written in everyday language, identify patterns and help carry out work that previously required someone to read, understand and respond to information manually. For a business with 30 or 40 employees, even relatively small improvements to everyday tasks can make a worthwhile difference. When those improvements apply to work that grows with the business, the benefits can grow too.
But solving customer problems isn't the only reason software companies are adding AI. They also need to compete. If their competitors are announcing AI-powered features, there's pressure to offer something similar. AI can help attract customers, justify higher subscription prices or create new revenue through usage-based charges. And, of course, providing AI services costs software companies money too.
These reasons aren't necessarily in conflict. A software company can introduce something genuinely valuable for its customers while also improving its own business. There's nothing wrong with that.
An impressive demonstration or description of a new AI feature doesn't tell you whether the feature will be useful in your organisation. It might solve a problem your staff don't have, duplicate something your existing software already does, or require more work to set up and manage than the time it saves. And even when it works well, it might cost more than the benefit it delivers. The question isn't whether the AI is clever. It's whether it makes a worthwhile difference to your business.
Before you switch it on, find out what's changed
This is something I think businesses need to pay more attention to. When your software provider announces a new AI feature, it's tempting to focus on what it promises to do. But there are other questions worth asking before anyone starts using it.
What information can the AI access? Can it read customer records, emails, financial information or documents? Does it simply provide suggestions, or can it change information and take actions on behalf of your staff? These are the kinds of questions we're hearing from business owners, including the person I mentioned earlier. You should also check whether the feature is switched on automatically, whether you can disable it and whether the provider has changed its pricing, terms or information-handling arrangements.
Importantly, some changes may affect your business even if nobody uses the new AI feature. We've seen why this matters before. In 2023, changes to Zoom's terms of service raised concerns about whether customers' meeting information could be used to train AI models. The wording caused confusion and prompted public criticism and questions about customer privacy and trust. Zoom subsequently revised its terms and clarified that it does not use customer meeting audio, video, chat, screen-sharing content or other customer communications to train its own or third-party AI models.
The controversy centred on what the wording appeared to permit, not on evidence that Zoom had used customer meetings to train its AI models. It demonstrated how ambiguous wording in software terms can create uncertainty, undermine confidence and prompt a provider to publicly clarify how customer information is handled. There's also an important distinction: an AI service may need to process your information to complete a task, but that doesn't necessarily mean the information is being used to train an AI model.
The important thing is knowing what arrangements apply to the software your business actually uses, rather than making assumptions. Check the relevant privacy and security settings, including whether information may be retained or used for model training. Also check what the feature can access, whether it changes permissions or connections to other software, and whether existing client confidentiality obligations are affected. A setting that prevents model training does not, by itself, answer every question about information handling. For a professional services business handling confidential client information, this deserves particular attention.
What if the AI is genuinely useful, but expensive to run?
An AI feature might be genuinely useful and still be too expensive to justify. The important question isn't simply whether it works, but whether the benefits outweigh the costs when you start using it regularly.
Consider email management, for example. Many businesses have been dealing with the same problems for years. Messages arrive, staff read them, work out what needs to happen and then manually pass information along or take action. The rules built into Microsoft Outlook can automatically sort, move or forward emails when they match conditions you specify. But they're much less useful when software needs to understand what someone is asking for and work out what to do next.
Microsoft's Copilot Studio opens up new possibilities. Rather than relying only on predefined rules, it can help businesses build AI assistants that interpret requests expressed in different ways and support more complicated workflows. Through our work, we're showing organisations how to use these capabilities and helping them put them into practice, particularly for problems that conventional email rules haven't been able to solve.
That's exciting because we're talking about solving real business problems, not simply generating emails or summarising documents. But there's a catch. You need to understand what it will cost when people start using it every day.
Microsoft has different licensing and charging arrangements for Copilot Studio. Some eligible AI interactions are covered by Microsoft 365 Copilot licences, while other activities consume Copilot Credits. Unlike a familiar fixed monthly software subscription, these credits reflect how much the AI is used: businesses may buy capacity in advance or pay according to usage, depending on their arrangement. The costs depend on what the AI does and how the solution is set up. If your staff already have Microsoft 365 Copilot licences, don't assume they cover an AI assistant working through incoming emails automatically. Check how that activity would be charged. Something that works beautifully when you're testing it on ten emails might have very different costs when it's handling hundreds or thousands of messages each month.
Imagine introducing an AI workflow that saves your staff several hours of work each week. That could be excellent value, even with a noticeable monthly cost. But if you're paying substantial usage charges to process relatively simple messages, a different approach might be more economical. The technology isn't necessarily good or bad in either case; the question is whether the benefit justifies the running cost.
Before committing to a solution, work out how much you'll actually use it, what happens during busy periods and whether there are ways to monitor or control the charges. An AI tool can work extremely well and still not be worth what it costs to run.
So, should your business start using it?
I wouldn't automatically say yes, and I wouldn't automatically say no. I'd start by asking what problem the new feature is supposed to solve. Is it something your staff spend a lot of time doing? Could it help them serve customers better, reduce mistakes or handle work that currently takes too much effort?
Then look at whether the proposed solution fits how your business operates. For example, if you're considering an AI tool to help manage client enquiries, don't just test whether it can understand a few emails. Look at what happens when the message is unclear, when information is missing or when someone makes an unusual request. Think about who will check its work, what it can access and what happens when it gets something wrong.
And, of course, work out the cost. You don't need to roll it out to your whole organisation to find out whether it's useful. A limited trial with a small number of staff and some real examples will often tell you far more than a polished demonstration. You might decide to start using it, test it further, leave it for now or decline it altogether.
Any of those decisions could be right for your business, and none needs to be permanent. As capabilities, pricing and your needs change, something that doesn't make sense today may be worth revisiting.
Before enabling a new AI feature
Eight practical questions to help you decide whether a new AI feature is right for your business.
- Will it deliver business value? Identify the expected time savings, service improvements or other benefits.
- What information can it access? Check access to customer records, emails, documents and confidential information.
- What happens to our information? Check how data is processed, stored and shared, whether it's used for AI training, and what controls are available.
- What can it do, and what if it makes a mistake? Consider its ability to take actions, potential consequences and where human approval is needed.
- What's changed in our software? Check settings, permissions, integrations, terms and whether AI is enabled automatically.
- What if we choose not to use it? Can we disable it? Will existing functionality, costs or data handling change?
- What will it cost? Check subscriptions, usage charges and whether the benefits justify the cost.
- How will we know it's working safely and effectively? Test it on real work, assign responsibility and monitor results and risks.
The decision should be yours
When I think back to NoVacancy Sydney, what stays with me isn't just how many software companies were talking about AI. It's the question of how businesses should respond when AI becomes part of software they already rely on.
There are some fantastic opportunities emerging. AI is making it possible to tackle problems that have frustrated businesses for years. But that doesn't mean every new feature is worth switching on.
Just because your software provider has added AI doesn't mean your business needs to start using it. Make sure there's a good reason to do so.
Practical takeaway
Before switching on an AI feature, check what it can access, what it can do, whether your provider has changed its terms and what it will cost to use. If there is a worthwhile business need, test it on a small amount of real work first.
If you're unsure how a new AI feature might affect your business, talk to AIBoK. We can help you work through the questions before you decide.
Frequently asked questions
Should my business use AI features that come with our existing software?
Not necessarily. Start by understanding what the feature does and whether it solves a problem worth addressing. If it saves your staff time, improves customer service or makes useful work easier, it may be worth trying. But there's no reason to adopt something simply because your software provider has made it available.
How can I tell whether an AI feature will actually be useful?
Try it on real work rather than relying on the provider's demonstration. Compare how the task is currently done with what happens when you use the AI. Include the time staff spend checking results and correcting mistakes. The important question is whether the overall job gets done better, faster or more economically.
How is AI different from ordinary software automation?
Traditional software automation generally follows rules you've defined. AI can interpret information and respond to situations that don't fit neatly into those rules. For example, it may understand what someone is asking for in an email without needing particular keywords. That flexibility is useful, but it also means you need to check how reliably it works.
What should I watch out for when my software provider adds AI?
Check what information the AI can access, what it can do, whether it's switched on automatically and whether the terms or pricing have changed. Look for additional subscription fees, charges based on usage and any need for staff to check its work. If the AI handles confidential client information, understand how that information is processed and protected.
Can new AI features affect my business even if we don't use them?
Potentially, yes. Your software provider may change subscription arrangements, terms and conditions or information-handling practices when introducing AI. Not every change will affect your organisation, but it's worth checking rather than assuming nothing has changed because you haven't enabled the feature.
Will the AI use our business or customer information to train its models?
Not necessarily. An AI service may process information to perform a task without using it to train AI models. The arrangements vary between providers and sometimes between different plans or settings from the same provider. Check the applicable terms, what information is accessed or retained and what controls your business has.
Why might an AI feature cost more than I expect?
Some AI features involve usage-based charges. The cost may depend on how often the AI is used or how much work it performs, rather than simply how many employees have access. A solution that seems inexpensive during a trial could cost considerably more when used throughout your business. Estimate normal and peak usage, and understand how to monitor the charges before rolling it out.
Matthew Tati is a co-founder of AI Body of Knowledge, where he focuses on practical AI training, capability development and adoption. His background spans more than 20 years in technology, product leadership and project delivery, including work in the travel and tourism industry and service on the Board of Directors of the Australian Tourism Data Warehouse. He has also served as Chair of IoT Alliance Australia’s Innovation Workstream. His work focuses on helping organisations turn emerging technologies into useful, sustainable working practices.