Published on the 12/09/2023 | Written by Heather Wright
ChatGPT, legacy tech and hybrid cloud…
For Barfoot & Thompson, embracing ChatGPT was an easy decision – and it’s one that is on the road to saving the company 32,000 hours a year.
The 100-year-old New Zealand residential and commercial real estate and property management company had been looking for a technology to help with a key problem. Its diverse staff base of 2,800 people across Auckland and Northland speak over 54 languages. While that’s great for business, it also has its challenges – especially in putting together the volumes on advertising copy required for its thousands of property listings.
“We could choose not to use anything with AI or we could educate our people and be ready for the future.”
Anil Anna, Barfoot & Thompson IT manager, says “With our diverse workforce where English may not be their first language, we have always had a challenge of publishing the content on our website, especially in our property management space where we have people who are experts in managing properties, but may not be an expert in copywriting or writing an advertisement quickly for their property.”
Time is of the essence for property managers, and Barfoot & Thompson was looking for a way to help them save time – and free them up for dealing with customers.
“It had been in the back of our mind for a few years and we were looking for opportunities to improve things.”
ChatGPT would prove just that opportunity – and an award winning one at that for Barfoot.
The company recently won the REINZ Innovation Award for its AdCompose AI offering – a tool that can create original adcopy for property managers, as well as fine-tuning existing adcopy.
Barfoot launched AdCompose AI for its property management team – it’s the biggest property manager in New Zealand outside the government’s Kainga Ora – in early 2023. By April it expanded its use to all of its sales team
Anna says the offering has the potential, if used to full capacity, to save around 27,000 hours a year in sales management and a further 5,000 hours a year on the property management side. That’s based on just 50 percent of sales ads harnessing the technology, with the time taken to write an ad slashed from four down to one hour. In reality, the time taken to write an ad can be up to six hours, increasing the potential savings. On the property management side, Anna says the savings are based on shaving an hour per ad in writing time.
“We haven’t measured it as such to say exactly how much time we are saving but we know there is a saving because we are quite close to our people and they are very good at giving us feedback,” Anna says.
The offering is also providing a bottom-line standard for adcopy produced while adjusting tone to suit, and ensuring distinct representation of properties. Managers review the copy once it is written.
“Any time we save our property managers or sales people is time they can put into serving their customers better and we see that as an opportunity for us to step up and raise the bar in terms of offering excellent customer service to customers and potential customers.
“Plus any time we gain out of these tools is going to help bring some work-life balance as well, so it works on a personal front and also on the business front.”
But Anna is also clear the development of AdCompse AI wasn’t about harnessing the latest technology.
“We always look for a use case before we think about a technology. If you’re thinking about a technology and trying to apply the technology to your business, it may not work, or you might be adopting technology for the sake of adopting it,” he says.
“But when there is a use case, where you’re seeing a real problem that you are trying to solve and then look for solutions and you adopt a technology, it can work at multiple levels.”
AdCompose AI, delivered as a beta offering in three weeks, enables Barfoot’s team to provide as much information they have on a property into the app and set a tone for the ad.
“We capture all the information and put the question to ChatGPT in a format that can return the result for us.”
At this stage, the app harnesses ChatGPT data only.
“As part of our IT architecture we have a data layer which we see as a core asset layer and we are looking at another use case in our business.”
That use case, currently in proof of concept, would see customer’s visiting Barfoot’s website able to ask in simple terms for what they’re looking for – ‘I want a three bedroom place in Mt Eden within $1 million’, for example – rather than having to search on preset options. Site visitors could also ask to see all the open homes for two bedroom places in Mt Eden this Saturday or Sunday, or for properties with specific features.
“We are working on a couple of options,” Anna says. “We are exploring Google AI capability at as well at the moment.”
One selling point for Barfoot with Google AI is that it won’t consume Barfoot’s data and make it public.
“Any data we hold that is exclusive to us has value and will give us competitive advantage, so we don’t want that data to be available for anyone else to use,” Anna notes.
“Knowing that the data stays with us and the insights stay with us makes us more comfortable having such a tool in our environment. Knowing we can tap into the capability of AI, but not compromise the data and insights it can provide.”
It’s an issue many companies are grappling with – and one that has seen plenty of nervousness and some companies saying they’re holding off on any use for now.
Anna, however, says while the risks are there, Barfoot has been quick to see – and indeed to realise – the value of the technology.
“We did evaluate the risks and challenges, but we see the value of this capability and also what we are going to give out to get this capability.
The biggest risk is the data privacy and the data security.
“Those are the two things we considered and, I wouldn’t call it significant, but always you have this fear of the unknown – you don’t know what you don’t know, and you don’t know what risks or challenges this might bring.
“We did think about this and we worked out that either we could choose not to use anything with AI or we could educate our people on how best to use this capability that is available and at our fingertips today and be ready for the future.”
While Barfoot is using its own AdCompose AI, it’s also allowed staff to use ChatGPT in general, so that education extends throughout the team, from the IT teams to the wider workforce, who are regularly reminded about the need to be careful what data is entered into ChatGPT, for example when asking it to help write a response to an email which may contain private or confidential information. They’re also reminded equally about being careful about the information they get out of ChatGPT.
Beyond generative AI
Technology is one of Barfoot & Thompson’s five pillars, alongside people, customers, leadership and governance.
The 42-strong IT team sees itself as a complete IT shop, providing services from the service desk right through infrastructure, networking and security, DevOps, an API team, web team, data warehouse and business intelligence team and business analysts and quality assurance and solutions architects.”
Innovation has, and always will be, factored into budgets, Anna says, while operating in an agile environment emphasises flexibility, collaboration and iterative development.
“This structure allows the team to adapt and try new things, to improve processes and deliver value to stakeholders.”
“We have a strong technology capability in house and we are proud to say we have spent a lot of energy into retiring our legacy when a lot of businesses are struggling to handle legacy, but also to invest in new capability.”
Anna says as well as removing legacy, the company has been focused on ensuring it’s not building new legacy into its environment.
“We took the approach from an operational perspective of don’t build any new legacy, clear the legacy, and focus forward and look at how we can deliver value to our users.”
The company is also embracing another common trend at the moment: Going hybrid rather than all-in cloud.
Anna is blunt that while the company evaluates cloud options almost every year, most of the time the cloud solutions prove to be expensive.
“So we have a hybrid model where we have some in Azure and a lot on-prem.”
It has two data centres, one in Auckland and one in Hamilton.
“Everything is possible because we have a good team and good leadership. We run a very flat IT structure with two manager across the entire team, and they’re all self-managed teams. We serve the team by providing the direction of where the business is heading and what our business priorities are and the team delivers,” Anna says.
“The amount of satisfaction our IT team get when they know the right tool is in the hands of of our people and that they are making best use of it, outweighs anything you can offer to them,” Anna says.
