The smart colleague – AI agent supports internal sales at VAUDE

Compact AI project with tangible benefits

The AI colleague reports for duty

he AI agent was designed to support internal sales in two specific scenarios: 

• entering incoming orders into the ERP system 
• generating backorder lists for customers when required 

The project was planned as a 12-week pilot. During this time, BE-terna’s AI experts designed, developed, and implemented the solution. VAUDE employees continuously tested the agent by providing test emails and validating its responses. 

The AI agent went live in mid-September 2025, right on schedule. Since then, it has been supporting the internal sales team by preparing incoming requests for processing. 

To date:

• 71% of incoming orders are automatically identified and entered into the ERP system correctly
• remaining requests are automatically forwarded to employees when the agent recognizes that manual intervention is required
• backorder list inquiries are already handled entirely by the AI agent

How the AI agent works

Incoming inquiries vary greatly—from simple orders to image attachments from the VAUDE web shop or Excel lists in customer-specific formats. The AI agent must interpret all these formats and extract the relevant information. 

When a new email arrives in the monitored mailbox, the agent first analyzes the content and determines whether it contains a task it is allowed to perform. This requires understanding the context of the message rather than relying solely on keywords. For example, an email stating “I have a question about my order for item XY” should not be treated as an actual order. 

Once a valid task is identified, the agent extracts the relevant information such as item numbers, quantities, delivery dates, and customer details. This can be complex: at VAUDE, item numbers consist of product number, color, and size, which may sometimes be described only in text form. 

After extracting and compiling the necessary data, the agent imports the order into the ERP system using the EDI import mechanism of the BE-terna Fashion industry module. This ensures that standard validation checks—such as verifying customers or items—are automatically executed.

Finally, the AI agent prepares a response email for the customer.

Transparency and control are paramount

The AI agent performs all steps with full transparency. In a dedicated interface, it documents how each email was processed and which decisions were made. 

Employees can see: 

• which tasks the agent completed how data was extracted from the email 
• which actions were taken 
• whether any tasks still require manual processing 

As part of a human-in-the-loop approach, employees review the proposed actions and approve them before they are executed.

 “From a purely technical standpoint, it would of course be possible to automate the process without human confirmation,” explains Köhler. “But we want to gain experience here."

"Even with the intermediate step of approval, the agent provides us with enormous relief: instead of having to process every email manually, our colleagues receive a completely finished processing proposal that they simply have to confirm with a click. Humans remain in control, while the agent takes over the time-consuming work.”

Expansion possibilities for the future

The AI agent uses a set of different tools that enable it to extract data, analyze attachments, and formulate email responses. Its capabilities can be expanded by adding new tools for additional use cases. VAUDE and BE-terna are already working on refining existing tools and developing new ones for special scenarios.

“Our agent does not learn independently. This may sound surprising, but the agent works in our production system. Incorrect entries could have serious consequences, so we want to retain control over how tasks are processed,”

emphasizes Daniel Köhler.

Looking ahead 

“The first stage was a complete success, but our AI journey is far from over,” concludes Daniel Köhler. The success of the AI agent in internal sales has sparked interest in other departments. VAUDE and BE-terna are now exploring additional scenarios, such as supporting customer communication about repairs or sustainability questions. 

In the future, agents may also share tools or delegate tasks to each other, enabling more advanced automation across processes. VAUDE also plans to train employees so they can actively support the development and use of AI agents within their departments. 

“The basic idea behind our AI efforts is automation. We no longer think only in terms of efficiency gains, but want to relieve employees of repetitive tasks and have them performed by AI,"  emphasizes Köhler.  

"The value of human work does not lie in typing information into system masks. It lies in tasks that require real thinking and meaningful human interaction—such as building strong relationships with our customers and suppliers.”