'Packing Assist AI' that Proposes Optimal Packing Sizes Adds 'Inner Carton Support' and 'Exclusion Settings'
ROMS Inc. has implemented two new features, 'Inner Carton Support' and 'Exclusion Setting', into its 'Packing Assist AI'. This hybrid approach combines AI calculation with human rules to close the gap between theoretical optimization and practical logistics site operations.
📋 Article Processing Timeline
- 📰 Published: April 7, 2026 at 20:00
- 🔍 Collected: April 7, 2026 at 11:30
- 🤖 AI Analyzed: April 20, 2026 at 22:56 (323h 25m after Collected)
ROMS Inc. (Headquarters: Shinagawa-ku, Tokyo; CEO: Yosuke Maeno), a provider of automation solutions for logistics warehouses and factories, has newly implemented two features: "Inner Carton (inner box) Support" and "Exclusion Setting".
This update was implemented based on feedback from adopting companies to bridge the gap between theoretical optimal solutions and the actual business practices on site. By combining the powerful computational ability of AI with human pre-settings, it achieves highly practical packing instructions.
## Background
The greatest strength of AI in packing optimization lies in its ability to instantaneously derive the minimum volume from a vast combination of products. However, logistics sites have inherent rules that cannot be determined by computational logic alone, such as "treat 24 bottles as one box" or "do not bundle this product because it has a dedicated case."
If all of these were entrusted to the automatic learning of AI, there is a risk of incorrect learning (overfitting) occurring due to minute irregularities, which would disrupt site operations. Precisely because ROMS knows AI well, it chose the hybrid approach that is most stable in actual operation: "advanced calculations to AI, operational rule definitions to humans."
## **Update Feature Overview**
The features this time were developed based on feedback (e.g., sense of incongruity in packing results) from workers actually using the 'Packing Assist AI'. Rather than the AI becoming a black box and updating automatically, it assumes an operation where administrators catch the site's sense of incongruity and correctly modify the product information, thereby constantly improving the accuracy of the entire system.
### 1. Inner Carton (Inner Box) Automatic Conversion Support
When a single product (piece) reaches a specific quantity (e.g., 24 bottles), it is automatically recognized as an "inner box" for packing calculations.
The Packing Assist AI takes on the site's judgment of "if 24 bottles, one inner box; the remaining 1 bottle as a single item." It saves the worker the trouble of calculating and dramatically improves the accuracy of packing material selection.
### 2. Exclusion Setting
This is a setting to exclude large items already in shipping package form, or items not suitable for bundling with other items, from the calculation.
By explicitly stating "exclude from calculation" on the product master data, it prevents the AI from making unreasonable bundling proposals. This eliminates redos and hesitation on site.
## **Information on Demonstrations and Free Seminars**
We provide various opportunities for demonstrations so that you can experience the operability and introduction effects of 'Packing Assist AI'. In addition to regularly held online seminars and individual meetings, you can view demonstrations at the ROMS Heiwajima Lab (Ota-ku, Tokyo).
Contact: https://roms.inc/contact
ROMS Heiwajima Lab: https://roms.inc/contact
This update was implemented based on feedback from adopting companies to bridge the gap between theoretical optimal solutions and the actual business practices on site. By combining the powerful computational ability of AI with human pre-settings, it achieves highly practical packing instructions.
## Background
The greatest strength of AI in packing optimization lies in its ability to instantaneously derive the minimum volume from a vast combination of products. However, logistics sites have inherent rules that cannot be determined by computational logic alone, such as "treat 24 bottles as one box" or "do not bundle this product because it has a dedicated case."
If all of these were entrusted to the automatic learning of AI, there is a risk of incorrect learning (overfitting) occurring due to minute irregularities, which would disrupt site operations. Precisely because ROMS knows AI well, it chose the hybrid approach that is most stable in actual operation: "advanced calculations to AI, operational rule definitions to humans."
## **Update Feature Overview**
The features this time were developed based on feedback (e.g., sense of incongruity in packing results) from workers actually using the 'Packing Assist AI'. Rather than the AI becoming a black box and updating automatically, it assumes an operation where administrators catch the site's sense of incongruity and correctly modify the product information, thereby constantly improving the accuracy of the entire system.
### 1. Inner Carton (Inner Box) Automatic Conversion Support
When a single product (piece) reaches a specific quantity (e.g., 24 bottles), it is automatically recognized as an "inner box" for packing calculations.
The Packing Assist AI takes on the site's judgment of "if 24 bottles, one inner box; the remaining 1 bottle as a single item." It saves the worker the trouble of calculating and dramatically improves the accuracy of packing material selection.
### 2. Exclusion Setting
This is a setting to exclude large items already in shipping package form, or items not suitable for bundling with other items, from the calculation.
By explicitly stating "exclude from calculation" on the product master data, it prevents the AI from making unreasonable bundling proposals. This eliminates redos and hesitation on site.
## **Information on Demonstrations and Free Seminars**
We provide various opportunities for demonstrations so that you can experience the operability and introduction effects of 'Packing Assist AI'. In addition to regularly held online seminars and individual meetings, you can view demonstrations at the ROMS Heiwajima Lab (Ota-ku, Tokyo).
Contact: https://roms.inc/contact
ROMS Heiwajima Lab: https://roms.inc/contact