Order confirmations are among the most time-consuming and, at the same time, most economically critical routine tasks in operational procurement. Each item must be checked for quantity, delivery date, price, and specification, as unnoticed discrepancies directly impact margins, material availability, and customer deadlines. Traditional solutions like OCR, templates, or image-based recognition quickly reach their limits when supplier formats vary or units need to be converted.

Oliver Kosalla (Head of Purchasing at Anton Debatin), Denise Raupp (Operational Purchasing at Anton Debatin), Nico Bentenrieder (Founder at Tacto), and Jakob Hafner (Product Development at Tacto) will demonstrate in the webinar how Debatin replaced the manual order confirmation process with AI agents, thereby freeing up substantial operational capacity for strategic purchasing work.

Manual Order Confirmation Review: The Status Quo at Anton Debatin

‍

Anton Debatin is a globally leading manufacturer of security packaging, document pouches, medical bags, and customized packaging solutions. The company employs approximately 174 people, generates around 35 million Euros in revenue, and manages a purchasing volume of approximately 17 million Euros with one strategic and five operational buyers. Before automation, around 2,700 order confirmations were manually reviewed per year, distributed among approximately 330 active suppliers. Each order confirmation took an average of 10 minutes to process: reviewing the incoming OC, comparing it with the order in the ERP, clarifying discrepancies, and filing the document in the DMS. In total, 225 hours per year were tied up in a routine task, without providing strategic added value for purchasing.

Why Traditional OCR Approaches Reach Their Limits

‍

Existing market solutions for order confirmation review are based on three approaches, each with structural weaknesses. PDF extraction and OCR provide unstructured text without semantic assignment, tables are broken apart, and scanned OCs are barely legible. Template-based systems require a separate template per supplier, break immediately with format changes, and do not scale. Image-based ML models require extensive training data and a separate training for each supplier, without the content being professionally validated. At Debatin, additional challenges arose that these solutions could not address: quantity specifications in linear meters with a simultaneous price per kilogram, suppliers who confirm only calendar weeks instead of fixed dates, and substitution rules where specialized knowledge is lacking.

‍

Automated Order Confirmation Review with Tacto Intelligence

‍

Tacto's solution extracts incoming order confirmations regardless of format or supplier, without the need to maintain templates or supplier-specific forms. Tacto Intelligence performs an item-by-item comparison of quantity, price, delivery date, and article specifications with the order and prioritizes discrepancies for purchasing. Integration into the existing ERP process occurs in three steps: importing order, item, and supplier data from the ERP; setting up email forwarding for incoming OCs; and writing back automatically processed or corrected OCs as a standardized CSV into the ERP. According to Matthias Sorg, Head of IT, the ERP integration at Debatin was implemented within one to two days.

‍

80 Percent Straight-Through Processing Rate: Debatin Case Study

‍

At Debatin, automated order confirmation reviews are now the new standard. Approximately 80 percent of all incoming order confirmations are automatically processed, with only 20 percent, containing identified discrepancies, being forwarded to the buyer for decision. Concrete practical examples show what the AI detects that was previously overlooked: unconsidered discounts in orders that led to price differences between the order and the OC, or shifts of a few days in the delivery date that would have been missed during manual review.

‍

In its internal business case, Debatin calculates an impact of 25.6 days of efficiency gain per year, which corresponds to a 15 to 20 percent saving in working time in operational purchasing. Oliver Kosalla summarizes the qualitative effect: "Order confirmation review was one of the most monotonous parts of our day. The team immediately noticed that this effort is now gone. We once again have capacity for tasks that truly create value."

‍

Conclusion

‍

Anton Debatin's case study illustrates that automating order confirmation checks is far more than mere process optimization. The key is to combine semantic document understanding with the existing ERP database, so that procurement only intervenes where truly necessary. This frees up capacity for negotiations, supplier development, and strategic initiatives, rather than routine controls.

Oliver Kosalla and Denise Raupp from Anton Debatin discuss with Nico Bentenrieder and Jakob Hafner from Tacto how the packaging manufacturer transformed its approximately 2,700 order confirmations per year from a manual routine into an automated process. They will show why traditional OCR and template-based approaches fail in practice, how Tacto Intelligence extracts order confirmations independently of the supplier and checks them against the order with line-item accuracy, and the concrete effects achieved by an 80 percent straight-through processing rate in operational purchasing.