AI training in transport and logistics

Select tasks for participants

A freight forwarder preparing a response to an inquiry needs different exercises than someone compiling documents for settlement. Therefore, before the training, we select tasks for the roles present in the group. For freight forwarders, the program can be based on extracting conditions from correspondence and preparing questions about missing data. For dispatchers, it would be useful to recreate current agreements and describe open issues for the next shift.

The settlement team can work on comparing read fields with the document and marking discrepancies to be explained. The goal of such an exercise is the ability to check the result. The decision to use the document is made by the employee according to the company's rules. During the session, there is no need to connect the tool with the accounting system or TMS.

Newcomers need time to prepare the first instruction, add allowed material, and improve the response. With people who already use ChatGPT or Claude, we can move on to more complex cases: later changes in correspondence conditions or conflicting data in attachments. The level of participants is determined before proposing the program.

Exercise: What is missing from the response to the inquiry

A sample training material may include a fictional email with the loading location, date, and cargo weight, but without pallet dimensions. The participant prepares a data summary for the freight forwarder and a question to the client with the help of AI. They check whether the response includes the dimensions the sender did not provide, and whether the tool correctly transferred the date.

The instructor discusses a specific part of the work and asks for a correction. Then the participant receives a further message that changes the loading date. They update the summary and point out the message that is the basis for the change. This allows us to see whether the participant can independently return to the source, even when the previous response was correct.

The exercise can be expanded to include correspondence in the language used by the team. Then the control also includes whether the translation preserves the conditions of the inquiry. The valuation of the transport and acceptance of the order remain outside this example. A separate article describes working with AI when passing on changes from dispatchers.

Preparing the company for the session

Before the training, we determine the available accounts and the rules for working with materials. If the company is just choosing a tool, it can order ChatGPT preparation for the team or Claude implementation earlier. Configuration of the environment and purchase of licenses are a separate scope. Exercises can start with fictional data, without connecting the company's email.

The supervisor indicates the tasks and the people who will participate in the session. Together, we determine whether we can use appropriately prepared company materials. Before sharing correspondence, it is necessary to agree on which information can reach the selected tool and who is responsible for preparing the files.

During the collaborative work, time is needed for participants to complete the exercise without simultaneously handling deliveries. Repeating the training for the next shift, shorter meetings, and discussion of work after the session require consideration in the order. We do not assume that one day will cover all roles and all shifts.

How to assess whether the team can work independently

At the end of the agreed program, it is worth allocating time for a new matter that the instructor has not previously solved with the group. The participant should prepare a response, point out the sources of important data, and name the information that is still missing. When assessing the response, we primarily check its consistency with the current agreements in the correspondence.

In further work, the supervisor may check a sample of similar tasks: whether employees return to source messages, correct added details, and know when to pass the matter for clarification. Such observation will help choose the topic of the next exercises.

If after the training it is still necessary to manually copy data between the box and TMS, it is possible to consider separately automating work in transport. Training and integration solve different parts of this task; the scope of system building is determined separately.

Ordering the training

Write which roles you want to train, in which languages the team works, and which tools they already use. It will be helpful to describe the task where it is currently difficult to assess the AI response. Based on this, we will discuss the program, group organization, and pricing. The initial rate is €1,150 net per day, according to the price list. Contact us to determine the scope for your company.

How much does the training cost and when does it pay off

The training starts at €1,150 net per day. We calculate the return on one process, without assumptions about the entire company.

According to a study published in Science (2023), the time for writing tasks with ChatGPT has decreased by 40%. Therefore, the scenario adopts a cautious weekly hour and checks it by measurement.