Mixing maps with mindfulness: what agentic AI means for automotive and logistics

Agentic AI Connected Driving Bird's Eye Traffic

Agentic AI doesn’t just respond to prompts but sets plans, makes decisions, carries out tasks on its own to achieve a goal.

"The other day, I asked AI for the best, warmest and cheapest place I could take my family on vacation in December," said Aleksandra Kovacevic, Senior Director of Responsible AI at HERE Technologies.

While her AI assistant did a great job finding what she needed—the right hotel, the best location and even the flights—it stopped short of actually booking them. “It doesn’t do it for you,” said Kovacevic. “It’d be amazing if it could handle everything.”

That simple example sums up what agentic AI is all about. It’s the next step in AI’s evolution: from tools that simply provide information to ones that can actually understand goals, work around real-world limits and takes action to help you reach them.

In industries like automotive, transportation and logistics, this isn’t just a small improvement. It’s a major leap in terms of how businesses make decisions, adapt and get results.

Agents of change

Today, location-based tasks still take a lot of manual effort. Say you want to find a peaceful café with outdoor seating and an EV charger halfway along your route.

You’d have to piece that together yourself: use one tool to calculate the halfway point, another to look up cafés, then check which ones have chargers and finally read reviews to see if they’re actually peaceful. That last part, especially, is something current systems aren’t great at understanding.

In an agentic world, you can just say what you need in natural language, and the AI takes care of the hard part. Kovacevic gave a more industrial example: “Re-route all trucks over 7.5 tons to avoid low-clearance bridges, prioritize EV-friendly stops and anticipate high congestion near port access roads at peak hours.”

That’s one request packed with rules, priorities and predictions. In order to set up a system to handle something like that now means writing out detailed instructions for every possible situation. Agentic AI changes that. It can build and adjust a plan on the spot, shaping its approach to fit the task at hand.

Bridging data and decisions

Many companies are eager to use agentic AI, but they are running into a critical challenge: complexity. Today’s AI systems often connect to company data through standard tools and APIs, which works well for simple tasks, such as fetching a report or answering a routine question.

But when it comes to domain-specific problems, like rerouting delivery fleets or managing supply chains, these approaches often fall short. The problem isn’t that the AI lacks access to data. It’s that it doesn’t have the structured context or reasoning capabilities to make sense of it accurately. Without that, the system can misinterpret information or generate plausible-sounding but incorrect actions ie, what are commonly called hallucinations.

As Kovacevic put it: “Large language models are excellent at working with language. But that doesn’t mean they understand the world or how complex systems behave.”

In other words, they can sound smart, but that doesn’t mean they understand operational constraints like vehicle height limits, legal driver breaks or how traffic congestion near ports affects real delivery times. Without deeper integration with structured systems and domain rules, even the best AI can’t yet move from generating insights to reliably taking action.

And in industries like logistics, there’s no room for guesswork.

ship shipping container transport transportation logistics port supply chain

Moment of truth

Companies in the automotive and logistics sectors need to prepare for this agentic future. Kovacevic’s prescription? A deliberate, step-by-step transformation focused on building trust and upskilling human operators.

The first step is encouraging real teamwork between people and AI. Instead of trying to automate everything right away, the goal should be to give human operators smart, AI-powered recommendations.

For instance, if a delivery truck gets stuck in traffic, the AI could analyze real-time data and suggest the best next move.

“It should recommend an action,” said Kovacevic. “But you, as the operator, should always double-check it. The agent doesn’t have the same judgment or experience that you do.”

This kind of “human-in-the-loop” setup is key. It lets the AI do the heavy data work while people make the final call, using their experience to confirm or adjust what the system suggests. Over time, this builds trust and makes sure that valuable, hard-to-define human know-how isn’t lost.

The next step is to make that implicit knowledge explicit. As operators review and correct the AI’s suggestions, they’re essentially generating feedback that can be used to improve the system over time. Whether through manual tuning, feedback loops or future retraining, this process helps capture and formalize expert knowledge.

Kovacevic noted that operators will need to train the agents by “pointing out and labeling the things they get wrong, so the system can learn from them.” Over time, this feedback-and-refinement loop shifts the human role from manual execution to AI supervision and training.

HERE is uniquely positioned to support this transition. With advanced mapping and AI tools like HERE AI Assistant, companies can embed real-time location intelligence into everyday operations, improving decision-making, optimizing routes and reducing operational friction.

Work, reworked

Ultimately, this journey is about upskilling the workforce. Kovacevic pushed back on the idea that AI will simply replace jobs. Instead, she believes it will redefine them. "AI will not take over your job," she said, "however, if you do not know how to work with AI, someone else will."

For complex industries like logistics, full automation might not be a near-term reality.

"Logistics is quite intricate, making it unlikely for agents to fully take over just yet," said Kovacevic. The path forward is one of gradual transformation, where humans and AI agents work together to create more resilient and efficient systems.

This is just the beginning of our exploration into the world of agentic AI. In our next post, we’ll take a closer look at the location technologies powering these systems and how HERE is helping build the infrastructure behind this new generation of AI-driven capabilities.