At the 2025 Australasian Fleet Education and Leadership Summit, Mike Branch, Geotab’s Vice President of Data & Analytics and Board Member, delivered a compelling keynote that unpacked the impact of artificial intelligence (AI) on fleet management. With energy, humour, and insight, Branch laid out how AI is not just influencing the sector—it’s completely reshaping it.
“This is my very first time in Australia,” Branch began, breaking the ice with an apology: “I must say, I’m very, terribly, terribly sorry to have brought the Canadian weather here.” But it wasn’t long before the conversation shifted from the skies to the cloud, where AI is fundamentally transforming the way fleets operate.
AI’s Rapid Trajectory
Branch illustrated the astonishing speed of AI advancement. “Back in 2019, if you asked GPT-2… to tell a funny story about a mischievous cat… it did an okay job,” he said. “Fast forward to 2025… you get a beautiful image created at the same time.” He reinforced the pace with examples in text, images, and video, showcasing how “only a couple of years” separate underwhelming outputs from stunning, photorealistic results.
He noted that 71% of companies globally are now using generative AI, double from the previous year. “We are truly in a revolution,” Branch declared. “And no joke about that.”
Drawing comparisons to the Industrial Revolution, he argued that AI’s breakthrough moment was the launch of ChatGPT. “You need breakthrough innovation, mass adoption, economic disruption, public awareness, and global impact,” he explained. “We, my friends, are absolutely at the start of this AI revolution.”
The AI Maturity Model for Fleets
One of the standout moments in Branch’s presentation was his analogy of AI’s development in fleet management to human growth stages—from infancy to adulthood. In the beginning, fleet systems are merely observational. “You see GPS data coming in, you see accelerometer data coming in… you’re just in the see phase.”
Next comes the toddler stage: alerting based on set rules like speeding or harsh braking. As systems mature, they enter early childhood, asking “why” through basic analytics and early machine learning. “We developed this machine learning model that would compare how all the police vehicles would cluster together, all of the utility vehicles… leveraging early machine learning.”
The teenage years bring foresight. “It’s about making predictions,” he said, referencing Geotab’s work in predictive maintenance and safety modelling. “Instead of having a 90 out of 100 [driver scorecard]… you get a probability of collision in the next 100,000 miles.”
Eventually, the system matures into “managed execution” and finally “automated actions,” where AI agents take over routine decisions. “This is the really exciting stuff, and we’re just at the precipice of a lot of this happening.”
Agents Are the Future
Branch spent a significant portion of his talk demystifying the concept of AI agents—autonomous systems that observe, plan, act, and evaluate.
“An agent is something that can perceive the world around it. It can make a plan of action, it can think, and then it takes action on something,” he explained. Geotab’s own ACE platform is already enabling this, allowing users to ask natural-language questions like, “Which vehicles in the fleet have the highest risk of collision?” and receive data-backed answers.
He painted a future where agents handle everything from driver coaching to rewards and insurance reporting—autonomously. “It contacts the rewards agent… the safety agent calls the insurance agent, make sure I’m okay, calls the police agent. Everything happens all autonomously in this perfect utopian situation.”
But he also issued a warning: “If you have garbage data, you’re going to get terrible, terrible insights multiplied by a million.” Bad data could cause systems to misread driver behaviour, fail to reward correctly, and even botch emergency responses.
Building Trust in AI
Trust, Branch said, is as crucial as innovation. “You need to have transparency… Whatever AI is doing has to tell you,” he stated. Privacy, consistency, and reliability are also critical, especially as AI systems can “hallucinate”—delivering inconsistent answers to the same question.
“You don’t want three different answers to the same question. You want the same data,” he emphasised. To mitigate these risks, Geotab runs rigorous testing known as “red teaming” to catch vulnerabilities and edge cases.
He also underscored the importance of clearly defined goals. “You might say, okay, I’m trying to improve utilisation. Well, what is utilisation? Is it power driven? Distance driven?” Vague objectives can lead AI agents astray, so fleet managers must be precise.
The Road Ahead
Branch ended his session with optimism and a challenge to the audience. “We are in an age right now where things are just exploding,” he said. “If we can do this right together… as architects of trust in AI… we really can unlock what’s next.”
And as for the future?
“I really do wonder what we think our ‘remember when’ moment will be,” he mused—perhaps a world where driver scorecards, traffic lights, and manual data analysis are relics of the past.
