As logistics businesses learn the benefits of AI, “vibe-coding”, agentic AI, and more optimization tools, don’t get carried away by potential upside without evaluating the risk.
Analysis of nearly 80 publicly traded software companies finds that businesses in operationally critical industries (healthcare, cybersecurity, financial services, and logistics) are the most cautious about AI adoption, and for good reason. The cost of an AI error in these sectors isn't an inconvenience. It's a breach of customer data, a missed delivery window, a failed compliance obligation, or a reputational incident that takes years to recover from.
For last-mile logistics operators, this tension between AI's potential and the trust required to deploy it is playing out right now. The platforms you choose, and how seriously they take security and responsible AI development, will shape your operational resilience for years to come.
AI is promising new capabilities for teams that are already stretched thin
Logistics teams, dispatchers, and routers are all keen to adopt technology that makes their job easier. Agentic AI is a huge step for complex processes being automated and routes being dynamically re-routed in real time in response to live conditions such as traffic or a missed delivery.
But as AI tools are being built faster and faster, logistics companies need to slow down to evaluate safety. Using AI tools to generate apps for example, completely skips security infrastructure. Giving internal teams access to vibe coding tools might solve some of your process inefficiencies, but completely open you up to new vulnerabilities.
Data risks for logistics companies
An AI system operating autonomously, with access to fleet data, driver communications, and customer records, is a more complex attack surface than a passive analytics dashboard.
Here are just some of the potential downsides to unsupervised or less secure AI tools:
- Customer personally identifying information (PII) such as addresses, contact details, and names being leaked
- Your list of customers, territories, and business information being leaked
- Driver PII, including shifts, locations, names, and vehicle information being leaked
- Backdoor access to critical systems such as billing, container tracking, and warehouse or transport management systems can be exploited
These might sound hypothetical and worst case scenarios, but they’re very real. Here are some very recent examples:
- In late 2025, Asahi Japan’s logistics systems were attacked and as a result, customer orders had to manually be taken via phone calls
- In 2024, Transport for London was attacked and customer data (including banking information!) was accessed
- Toll Group was attacked twice in 2020, affecting billing, container tracking, and port logistics with an estimated damage of $300million
This is why established software tools are worth the investment. The procurement process can be frustrating, but worthwhile when the tool’s cybersecurity capabilities are evaluated.
Adiona is leading the way in cutting edge AI for logistics and cybersecurity
Adiona takes security seriously, and actively avoids ingesting PII to prevent data leaks and regularly gets third party penetration tests performed on our platform to prevent vulnerabilities from being exploited.
In our most recent penetration test, we received an A+ rating.
Adiona's FlexOps platform is purpose-built for the realities of last-mile logistics, not adapted from a generic optimisation engine but designed from the ground up to handle the complexity of real delivery operations. That means dynamic, multi-constraint route optimization that accounts for driver hours, vehicle capacity, customer time windows, traffic conditions, and even EV-specific variables like charge levels, range anxiety, and battery degradation modelling.
Adiona's Fleet Simulator adds another dimension – the ability to model EV transition scenarios, test charging infrastructure assumptions, and stress-test route coverage before committing to fleet or infrastructure investment. In an environment where operators are making expensive, long-horizon decisions about fleet electrification, having a simulation layer built on verified, real-world data is a meaningful differentiator.
What to Ask Your Route Optimisation Provider
As AI capability becomes table stakes, the differentiating questions for operators evaluating logistics software are increasingly about trust, security, and long-term fit. Consider asking:
- Has your platform undergone third-party penetration testing, and are you willing to share the outcome?
- How are your APIs secured, and what happens if an integration point is compromised?
- How does your AI make decisions and can those decisions be explained and audited?
- What operational data underpins your AI models, and how long have they been in production?
- How is your platform positioned for agentic AI and what safeguards exist around autonomous decision-making?
- Are you working toward or compliant with recognised security frameworks such as ISO 27001 or SOC 2?
Any platform positioning itself as an AI leader in logistics should be able to answer these questions with specifics, not generalities.