AI for the Email-Based Ordering Process

Our Artificial Intelligence automates the processing of unstructured email orders for any ERP system.

Time-Consuming Email Orders

Many wholesalers face the challenge that their customers continue to submit orders via email. The actual order may be contained in the text of the email, attached as a PDF, (Excel) spreadsheet, or even as a photo or scan. Manually transferring each individual order into the in-house ERP system is extremely time-consuming and tedious for sales staff—especially when customers use their own product names, which then have to be manually matched to the company’s own assortment. If the entire email correspondence or the sales history of the respective customer also has to be consulted, there is often little time left for the core tasks on which the sales staff would rather focus. This also affects the efficiency of order processing.

The Central Problem

A structured ordering process (for example, via a dedicated online ordering platform) would reduce the effort on the part of wholesalers but would significantly increase it on the customer side. The fundamental question is therefore:How can the efficiency of the ordering process be significantly increased without losing the familiar flexibility that customers appreciate when ordering?The manual effort is so complex that it could not or only partially be automated until now. However, with the latest developments in the field of Artificial Intelligence, this has changed.

The Solution: AI Order Automation

• Recognizing orders ✓
• Understanding technical terms and abbreviations ✓
• Understanding context (email history, queries, etc.) ✓
• Supporting common order formats (continuous text, PDF, Excel, etc.) ✓
• Unambiguous assignment of customers based on the customer database ✓
• Inclusion of customer-specific sales history ✓
• Seamless integration into almost any common email system (Outlook, Gmail, etc.) ✓
• Seamless integration into almost any ERP system ✓
• Data protection compliant ✓

Order Automation with AI – Intelligent and Seamlessly Integrated

Our AI-powered software ensures that sales employees can focus on customer service instead of spending time on order intake—without changing the ordering process from the customer’s perspective. Our system combines all relevant components of the ordering process: from order receipt via email, through the recognition of ordered items, to systematic filing in the ERP system. We use Artificial Intelligence to capture and process the order information—regardless of the format in which the order is placed. By integrating all important data sources, our AI adapts individually to each customer: frequently used technical terms and abbreviations are learned, and the sales history of individual customers is taken into account. The previous email correspondence between sales and customers is also analyzed to consider information from the context.

Our AI solution integrates seamlessly into existing workflows and works “invisibly” in the background. Neither customers nor wholesalers have to adjust their processes. We support common software applications: whether Outlook, Gmail, or another email program is used, and regardless of which ERP system is employed—our AI is universally applicable.

AI Order Automation in Detail

1. Order Receipt
   At the beginning of the process is the receipt of the order. Our system integrates seamlessly into the existing ordering process. In the case of an order via email, this means that the system gains access to the internal mailbox and can thus recognize and further process customer emails. If desired, order emails can also be specifically forwarded to a dedicated email address instead of granting the system access to the entire mailbox. By connecting to the ERP system and the contact addresses stored there, each email (as well as the order later automatically created) can be clearly assigned to the corresponding customer without manual assignment. At the same time, data protection is ensured because no internal or irrelevant email communication is processed.
2. Intelligent Order Recognition
   Subsequently, the email content is analyzed using Artificial Intelligence. This determines whether the email is actually an order—or another type of message such as inquiries, complaints, or product queries. Thanks to the use of state-of-the-art AI language models, order recognition is precise and reliable and goes far beyond simple recognition of certain keywords. The context of previous customer messages is also included in the analysis.
3. Recognition of Ordered Items
   The heart of order automation is the precise recognition of the ordered items, including the respective package sizes and order quantities. This step presents a major challenge for conventional systems because orders can not only be received in various file formats (free text, PDF, Excel, etc.) but also differ in their structure. Some customers specify the total order quantity per item, others provide a specific package size, while others may only indicate the number of units ordered. Additionally, customers’ item descriptions often deviate from the exact names in the company’s own assortment. Our order automation solves these problems by:
   
   • Utilizing the power of current AI models to reliably understand even complex orders while considering the context.
   • Processing visual file formats (PDF, JPEG/PNG) effortlessly thanks to intelligent text recognition.
   • Being directly connected to the ERP system and therefore able to access all relevant customer-specific information (e.g., the order history) during processing.
4. Creating the Ordered Items as a Structured Order
   The recognized order is finally automatically transferred to the ERP system. Each automatically created order is marked as AI-generated to allow for a final review before release. Additionally, the responsible employees are informed via email about the completed processing status to make the workflow transparent and efficient.

A Flexible & Scalable System

Our AI order automation is modularly built so that the required interfaces to the email, CRM, or ERP system can be easily connected. Likewise, additional systems and data sources can be integrated as needed, such as querying the stock level at the time of order to be able to switch to equivalent substitute products. The information to be extracted can also be flexibly adapted to each application: are additional details such as the delivery date specified by the customer or the delivery address required besides the ordered items? Thanks to the use of generalist AI language models, this is no problem and requires only minimal adjustments. Furthermore, our entire system scales depending on the current order load. From a few orders per week to several per hour: the required resources are dynamically activated or deactivated to enable immediate processing while keeping operating costs minimal.