Where does the real bottleneck in transport planning lie?
In most transport companies, orders arrive via email, PDF attachments, customer portals, and sometimes WhatsApp. All of that information then has to be entered manually into the TMS: loading address, delivery address, weight, reference number, requested delivery date, special instructions. At twenty orders per day, that is manageable. At a hundred orders, it becomes a full-time task that is also prone to errors. As a result, planners spend a large part of their day re-entering data rather than doing actual planning: capacity management, customer contact, and resolving conflicts. That is precisely the gap where AI is useful, not as a replacement for the planner, but as a handler of the preparatory work.
How does AI automation of order entry work in practice?
AI automation of order entry is built on document recognition and language processing trained on the structure of transport orders. An incoming email with a PDF attachment is processed automatically. The system identifies the relevant fields: loading and delivery location, date, weight, reference, special instructions. These are converted into a structured order object and prepared in the TMS. The planner sees a proposal with a confidence indicator. When confidence is high, they approve it with a single click. When uncertain, they make adjustments. The recognition rules are not automatically updated based on corrections; instead, they are periodically maintained and refined by the team. That is a deliberate choice: predictability over autonomous adjustments that cannot be accounted for.
What changes for the planner after automation?
Automated order entry does not replace the planner, but it shifts their work. What disappears: manual re-entry, searching for attachments, counting lines on a packing slip, and double-checking whether a reference number is correct. What remains: reviewing exceptions, coordinating with customers about deviations, working through capacity puzzles, and making decisions when requests conflict. Those are precisely the tasks where an experienced planner adds value. In projects we deliver, planners almost always describe this as the biggest shift: from typing to thinking. The workload does not necessarily decrease, but the quality of the work improves.
When does automating transport planning software with AI not make sense?
Honest advice is appropriate here. If orders vary widely in format, language, and structure and follow no recognizable pattern, recognition accuracy will be too low to add value. If the TMS itself cannot be reached via an API or structured input, the integration falls back on fragile screen automation, and that is not a solid foundation. If volume is low enough that the planner handles it comfortably without pressure, the business case is thin. Start with a measurement: how many orders per day, how many unique customer formats, how consistent is the structure of incoming messages? Those three questions determine whether automation makes sense. Anyone who answers them honestly will know within ten minutes whether to look further or not.
An AI Worker on top of the existing TMS, or a new core system?
If the existing TMS works well enough and the entry side is the only problem, an AI Worker is the shortest path. It reads incoming messages, processes documents, and feeds the TMS without touching the core system. Shorter lead time, lower initial investment, and the planner notices the difference immediately. But if the TMS itself is outdated, too rigid for the current customer mix, or a constraint on growth, rebuilding the core system is sometimes the smarter move. A modern, custom-built core with AI built in offers more flexibility than a layer on top of a system that has already reached its limits. Bonsai does both, and the conversation always starts with the same question: where is the real bottleneck?