Mike Baxter is Triquestra’s CTO. Here he talks about how he sees the future of AI unfolding for retailers.
The surge in interest in AI, especially in the past year, has retailers asking how they can incorporate it into their businesses for maximum effect. And at a time when 65% of Asia Pacific CEOs anticipate that AI will significantly enhance their organisation’s efficiency, it makes sense to think clearly about how to implement AI tools in a way that enhances productivity, streamlines processes and adds to revenue growth.
AI has the potential to revolutionise the retail sector, including when it comes to customer-facing aspects such as personalisation, as well as the ‘under the bonnet’ tasks involved in inventory management, the bedrock of any retail business.
Global retailers like Zara and Amazon are already using AI to monitor inventory in real time, dynamically allocate stock based on local trends and predict demand for millions of its products in just seconds.
As someone who spends a lot of time thinking about technology and how to integrate it into the ways Triquestra and our customers do business, I’ve got some thoughts on how the future of AI might play out and how retailers can get ready to make the most of it.
The next next big thing
Ten years ago the term on everyone’s lips was ‘cloud’, and we were all asking how we could incorporate cloud technology into our business for best effect. Now that AI is the next next big thing, it’s worth pointing out a few important differences between these two leaps forward.
For one thing, AI is a simpler proposition. Whereas cloud implementation required significant IT infrastructure investment and configuration, AI is already integrated into many existing tools, and most people are familiar with the best-known ones, like ChatGPT. That alone makes AI more accessible and easier to understand compared to the initial stages of cloud adoption.
That doesn’t mean it’s all plain sailing, though. It’s important to recognise that AI comes with its own challenges and pitfalls that you need to keep in mind when integrating it into your business.
Start with being clear about what you’re trying to achieve. Learn from the mistake that some businesses made during cloud implementation when they were carried away by the hype instead of clearly defining their goals. Start with something small, solve it, learn from it and then repeat it.
Get the inputs right
When people talk about AI, they’re really talking about a few different things all working together. If we think about demand forecasting, there’s a layer of machine learning that trains models on inputs like algorithms and data and then produces predictions of future customer behaviour. AI enhances this process by automating the analysis and selection of the best algorithms and data.
But AI’s effectiveness relies on the quality of the data and the instructions it’s given. In other words, it can only offer quality analysis if it has quality information to work with. Proper data management and preparation are essential for achieving accurate and useful AI outputs.
As Triquestra’s CEO Greg Cantlon said recently, ‘The retailers who will win the AI race aren’t necessarily those jumping on every new AI trend — they’re the ones building strong foundations with systems that make their data easily accessible and actionable.’
It’s also vital to ask the right questions. High-quality, tailored prompts that give precise context will return more helpful, accurate results than vague queries that take a shot in the dark. Without quality prompts, language models will give different results even with the same data.
Let’s say you’re a furniture retailer who wants to know about upcoming demand for outdoor furniture over summer. You’ve opened more urban stores in the past 12 months and warmer than expected temperatures and lower rainfall are expected. By using this data and targeting your prompts to querying sun umbrellas instead of all categories of outdoor furniture, you learn that increased sales in the range of 10% are likely and can plan accordingly.
Be willing to pay
Poor-quality prompts and data don’t just return variable results, they can also waste time and cost money, especially if you are paying to use an AI tool.
AI pricing models are based on tokens, where each token represents a chunk of data and the cost depends on the volume of data processed and the complexity of the prompts. Therefore, targeted prompting is a good way to optimise cost efficiency, whereas generic questions and bloated data sets can incur unwanted expense.
Still, it’s worth investing in paid tools, given that they offer better efficiency, especially when using targeted prompts. They also provide the benefit of enhanced data security and retention, whereas free tools can scoop up your data to train models for other users.
When it comes to choosing an AI tool, be willing to pay.
Prepare for rapid change
One of the standout aspects of AI technology is how rapidly it’s developing and changing. So, for example, language models are starting to specialise in different areas, such as writing software code, analysing customer sentiment, and personalisation. This specialisation could lead to more accurate and relevant outputs.
The key, though, will remain quality inputs. As AI advances, the need to interact with it in an intelligent, informed way will increase rather than decrease. That means developing people who understand your business and can interact with AI in a way that gets the best, most useful results.
Even at a time of rapid technological change, the human factor will be crucial to your success.
So will data quality and integrity. And that’s where Infinity comes in. As Greg said, ‘The question isn’t if your business will need to adapt, but how quickly you can do so when the time comes. The key is to choose systems built on robust APIs and data accessibility. Your point-of-sale system shouldn’t just handle today’s transactions — it should be your launchpad for tomorrow’s AI innovations.’
Want help to manage your data for best results?
If you’re looking for help to build a solid foundation for data accessibility, get in touch. We’d love to help you prepare for the future of retail.
