Australian firm AirHealth has revealed how it is using on-prem cloud storage to feed AI and create real-time pollen forecasts, in a gamechanger for asthma and allergy sufferers.
Addressing a media briefing in Sydney this week, Synology’s ANZ country manager, Jacqueline Graciella cited AirHealth as one example of how Australian businesses are tackling two major “structural shifts” currently at work: data readiness and digital sovereignty.
Garciella cited MIT research suggesting as little as 5 per cent of companies are generating meaningful returns on their investments in AI, despite local companies having almost universal uptake of the technology in some capacity.
Driving this ROI gap are a lack of oversight into how AI accesses unstructured data – including across documents, images, videos, communication logs, and institutional knowledge – and concerns around data security and accessibility.
“That information is reflecting how organisations actually operate, however most AI initiatives focus on structured data and, narrowly, automation. The real enterprise value sits in the unstructured data, but most organisations are simply not architected to activate it at scale,” Garciella said.
“Without the right data foundation, AI still remains experimental and not transformational.”
According to Garciella, AI readiness is not defined by a single LLM model or product, but rather a layered capability. Synology uses a four-layer approach:
- Ingestion: eliminating data silos and consolidating assorted data streams into a unified foundation.
- Infrastructure: ensuring data inputs remain accessible, secure and scalable.
- Semantic capability: transforming raw content into useable data that LLMs are capable of understanding.
- Agentic workflow: leveraging data inputs into AI to facilitate outcomes including automation and predictions.
“This is not a retreat from the cloud, but a rebalancing,” Garciella said.
Data readiness
On 21 November, 2016, a storm that swept through Melbourne would come to be known as the world’s largest thunderstorm asthma event. Respiratory-related presentations to hospital emergency departments soared 672 per cent, Victoria’s ambulance network was overwhelmed and 10 people ultimately died.
Since then, healthtech start-up AirHealth has pioneered a thunderstorm asthma forecasting service in Australia, using sensors to monitor pollen levels and make forecasts on air quality to help Australia’s 4.6 million hayfever suffers and attempt to reduce the 400-strong annual death toll from asthma.
Co-Founder Dr. Edwin Lampugnani said that in AirHealth’s early years, measuring air quality was a labour-intensive process involving manual daily collection of samples which were then analysed under a microscope to create a 24-hour average pollen count. However, new technology has enabled the company to substantially scale its operations and feed AI algorithms to generate more timely forecasts.
“We can deploy these systems all over Australia, we don’t need specialised personnel, and it means we go from having data that was collected only once a day down to 15-minute averages,” he said.
AirHealth’s sensors take 3D images of particles in flight, generating morphological fingerprints that AI uses to distinguish pollen from other particles.
With each device generating 6 terabytes of data annually, and strict government data security requirements to meet, AirHealth created a dedicated NAS hosted at its Victorian headquarters to store and activate its incoming data.
“That sounds really simple, but here’s the key – I’m a scientist, I’m not an IT infrastructure guy. Private cloud has enabled us to collect the data we need and for anyone to be able to manage it,” Lampugnani said.
The prominent aerobiologist labelled the transition as a “gamechanger” for AirHealth’s data collation and forecasting capabilities.
“We went from a system that involved manual retrieval to scalable automated transfer of data through Synology Drive; we went from months and even years of delay down to near real-time access to the raw data, and we are able to do that on-prem; we went from fragmented storage on consumer-grade hard drives to fully centralised one point where the data was all stored in a managed and structured way; and we went from at-risk data on consumer-grade hard drives to protected data that was appropriately backed up,” he said.
“We [also] managed to not employ more people just to look after a network system, and that meant that we could put the money that we would have spent on a network engineer back into the science to support more Australians.”
Operational survivability
Cybersecurity and IT solutions provider LEAP Strategies suggests that a DIY approach to technology and AI is very common in Australia, yet with mounting scrutiny over security, where that growing trove of data is hosted and accessed is more important than ever.
“There is a lot of noise out in the industry that we’re getting from clients about data sovereignty,” said CEO Grant Crough.
“It is data that sits in this country and is controlled by the laws of this country.”
He noted that any data hosted by a foreign service provider, even if it uses a local data centre for storage, means that a foreign entity has a stake in that data.
As well as the resultant exposure to foreign regulatory and geopolitical issues, cost becomes another challenge. With many offshore service providers hiking prices by double digit percentages – and subject to currency movements – continued access to stored data can become increasingly cost prohibitive.
Crough said that achieving data sovereignty, using on-prem hosting capabilities, means it is effectively free to access data, both now and in future, alongside far greater controls over access, security and disaster recovery.
“Not only is there a solution, there is a scalable solution. It’s not just flick a switch and do it tomorrow… but it is possible.”
