Canteen Australia’s pursuit of a greater good through tech

On in-house vs. partnerships: In the non-profit space, we have to be so responsible with every dollar our supporters give us. It’s always about ways to reallocate resources so we can create value while keeping costs at a similar level, or if the value we create makes it worthwhile to invest a little more. So it’s critical to have very trusted partners. And considering what we do, there are areas that are better to bring in house, so anything using data or technology in a way that’s going to impact our communities more directly. Then we could continue working with some of our trusted partners on some of the stuff that’s more removed, like our networks and key pieces of software. So what’s happening in the public-facing side is already integrated into what’s happening in the back office, which becomes a little harder when you’re having to work with lots of external parties that need to be brought together. But we’re a lot more efficient in how we use data, and we’ve now fully automated every pipeline. As a fundraising organization, you have to do so many little things to raise the resources to support young people. You need to work with a lot of different platforms and bring all that data in almost real time and create value from it. It felt critical for me to be able to create those capabilities in house.

On gen AI: Seeing how things were quickly emerging, we needed to look into ways we could start using it. I was very open to tell the team how we could use gen AI to make us more productive. So people were using Copilot very early on, for example. And that’s been helping them be a little more productive. But we haven’t removed anything, or said we’re going to replace a human with a regenerator. Instead, we’re helping people explore ways to become more efficient, which gives us more confidence in some of the work we do. We’ve also been exploring ways to help people get services where we may not be able to talk to them directly because it might be after hours. We did some research on people who care for cancer patients and they’re so busy the day so when they’re able to go online to check for things after hours, we were able to organize information and create models that show the data has been approved by us. But we still have to go through the process of building confidence, because the biggest fear people have is from AI providing inaccurate information.

On data security: Last year was big for cybersecurity, which helped unlock more resources to work on the space and make sure we keep information secure. One thing has been expanding the number of AI models we use to predict the best people for us to target for different campaigns. In the past, most non-profits had to send portions or entire databases to external suppliers to be able to run some of their own tools to decide whom to target. So it helped us accelerate some of that. Now we’re not sending data externally, apart from some very small data sets for specific purposes. And we’re able to do the rest of the work in house. We’ve also seen how we can use technology to mitigate a lot of risks that, in the past, have created opportunities for cyber attacks. Of course, we also have the human element to keep people more aware of things they need to need to watch out for. So for us, particularly because a good proportion of our staff are clinicians, it’s harder to have them think about the things they have to do to stay safe, including moving into hybrid working. So it’s been about updating all our tools so we can have more integrated ways to protect our users. We felt we needed to elevate some approaches we used to detect how much we were getting targeted. Before, only around three out of every 100 emails were filtered for spam, and now it’s gone up to about one in 10, so we’ve had to restrict access to some of our critical resources. With a little bit of monitoring, we get an alert whenever anyone is trying to download particular fields, or a particularly large file. We also talk to people just to understand the process they’re doing, and how we can help do it in a different way rather than downloading so much information about someone. I feel people are really on board.