Melbourne Airport has started leaning on agentic AI for incident response across its aerodrome and terminals, including in scenarios requiring rapid reaction.
The airport operator’s head of data analytics Irfan Khan told attendees at the Sydney Microsoft AI Tour that the company had started integrating agents with its SharePoint repositories in ways that allow staff to follow standard operating procedures (SOP) quickly.
It’s also using AI to save time preparing shift incident reports that are given to the airport operator’s senior leaders.
Khan said that the SOP documents contained airport policies that could apply to responding to everything from runway mishaps such as aircraft tyre bursts to passenger medical emergencies.
However, he said that, until now applying the information relied on humans that had experience with previous incidents or who had gained knowledge of the procedures in some other way.
“Historically, this has been information has been kept either in people’s heads or SharePoint but, when you have an event, you need to respond to it immediately – you can’t wait for someone to tell you what to do nor can you go to an SOP document to work or what to do,” Khan said.
“If you need an SOP for something, you need to know, ‘What should I do with this event?’.
“[Now], you simply ask the agent and we guarantee that you always get the latest information. If there’s a new SOP that was added or updated, you will always get the latest version of that.”
Melbourne Airport also has an offsite operations centre monitoring events across its facilities continuously.
Staff at the facility are required to prepare reports which is fed up to Melbourne Airport’s key leaders including its chief executive officer.
Khan said preparing the reports was often a chore for staff already grappling with fatigue following long shifts. Staff were now able to use AI agents to prepare them.
The agents, he said, also had given staff the ability to prepare reports with granular information specifically tailored for particular areas of the airport operator’s leadership team.
“It’s a very common occurrence that people often get these ad hoc questions or for the leadership to know an event has happened or they need to prepare for something, but the people who have to provide the information are very time poor. Agents help us a lot,” Khan explained.
Khan said that the airport is continuing to identify use cases for AI with demand frequently driven by staff sharing experiences internally.
However, he conceded that deployment had at times struck challenges. For instance, he said, the agents had been able to surface sensitive personal information inadvertently shared from OneDrive accounts.
Khan said that the situation prompted the airport to “step back” and start to reconsider its agents’ access permissions.
“Our agents that currently run on SharePoint are on sites and areas where we know for a fact if there’s anything sensitive in there it won’t be meant to have access to it. And that is, I think the biggest barrier that we currently have with having agents running everywhere, is that we need to govern,” Khan said.
