27/08/2026
零售從展店戰進入效率戰:熱食供應也該從多排一個人,改成多一套可複製流程
今日趨勢|2026年8月27日
台灣零售與服務業近期明顯從「展更多店」轉向「提升人效」。8月25日最新報導指出,超商與超市正透過自助結帳、電子標籤、行動支付等工具,把員工從結帳、價格更新與重複性作業中釋放出來;而今天8月27日下午登場的 ASUS EXPERTHUB 服務零售業論壇,也直接點出一個重點:企業數位轉型已經不是導入單一工具,而是要把營運流程、排班、銷售、會員與管理系統真正串接起來。
這個趨勢放到餐飲業來看,其實更明顯。
缺工時代真正的競爭,不一定是誰擁有最多機器,而是誰能把一套有效的流程,穩定複製到更多場域。
過去零售業競爭,常常看誰展店快、店數多、坪效高。但到了2026年,人的時間已經成為更珍貴的營運資源。自助結帳、電子標籤與行動支付的目的,並不是把人全部移除,而是減少員工花在重複工作的時間,讓人力可以轉去處理真正需要判斷、互動與服務的工作。
餐飲業也是如此。
如果企業、飯店或校園到了晚上10點,仍有20個人需要吃熱食,傳統做法通常會想到幾個方式:再多排一個人、延長廚房時間,或多設一個供餐櫃位。
但當場域從1個變成5個、10個,甚至20個據點時,如果每一個地方都繼續靠增加人力來支撐供餐,管理成本、人力調度與品質控制都會越來越困難。
運吉想解決的,就是這個問題:
能不能先把「一份熱食如何被保存、購買、加熱、取走」變成一套可以複製的流程?
今天用「起司抓餅」來說明。
起司抓餅很適合用來理解,為什麼不同食品不能全部使用同一種加熱方式。抓餅需要外層酥香,但起司與內餡又必須確實升溫。如果整份產品只靠單一微波,起司雖然容易快速融化,但餅皮可能因水分重新分布,失去原本應有的酥脆層次。
如果全部只使用蒸氣,餅體雖然能得到較多水分,但對於原本希望呈現酥香外皮的產品來說,蒸氣過多反而未必是理想選擇。
因此,運吉不會把所有產品都設定成同一個「微波三分鐘」。
我們會先用真正要販售的起司抓餅進行測試,再依產品實際狀態設定加熱程序。例如,可先透過微波協助起司、內餡與中心快速升溫,再搭配適當的紅外線加熱程序處理表層,讓外皮恢復較好的烘烤與酥香口感。
如果這道產品不需要蒸氣,就不必啟動蒸氣。
如果下一款換成肉包,可能需要的是微波加蒸氣;換成湯品,則可能要重新安排液體與固形物的加熱順序;換成白飯、燉飯或麵食,又會建立另一套參數。
這就是複合式加熱真正有意義的地方:
不是熱源越多越好,而是每一種食品只使用它真正需要的程序。
保存端也不必只有一種答案。
依食品本身的保存驗證、包裝方式、配送距離與實際銷售速度,可以從冷凍一路規劃到約 -8°C~-10°C 的低溫/微凍區間。
例如,中午銷量大的產品,可以提高補貨頻率與周轉速度;晚上或夜班使用的備用餐,則可以少量配置。某一棟辦公大樓一天需要20份,另一棟只需要8份,也不必全部採用相同庫存邏輯。
因此,運吉第一階段並不要求企業直接導入一整套複雜AI平台。
先用標準機落地。
中央廚房或食品品牌先完成產品製作、定量分裝、包裝與低溫配送;運吉設備則負責現場保存、自助選餐、支付、依餐點程序加熱與取餐。
先把最重要的營運問題跑出答案:
一天到底賣多少?
晚上哪個時間有人買?
起司抓餅與飯類,哪一種周轉比較快?
每個據點應該放多少?
多久需要補一次貨?
當第一台、第二台真正開始營運後,再依需求增加 AIoT、溫度紀錄、庫存管理、銷售統計、異常通知、補貨提醒與遠端管理。
這時候,管理者看到的就不只是「機器有沒有開」,而是可以逐步掌握更有價值的營運資料:
南科A棟晚上8點後平均銷售12份;
B棟週五需求下降;
某台設備只剩4份;
某款產品連續三天銷量增加;
某台設備溫度異常,需要檢查。
當這些資料回到中央廚房,下一批備餐量與配送安排就能更接近實際需求。
這也正是服務業與零售業數位化正在走的方向:
不是為了無人而無人,而是讓人不必一直重複做同一件事。
運吉可以從飯店、科技企業、校園、中央廚房、食品品牌,以及創業新品測試開始,先建立一個實際場域。
先讓一款起司抓餅跑起來。
再加入第二款、第三款產品。
場域增加後,再把庫存、銷售與 AIoT 接起來。
店不一定要開得更多。
把一套有效的供餐流程複製出去,也可以讓服務走得更遠。
English Version|Retail Is Shifting from Expansion to Efficiency: Hot-Meal Service Needs Repeatable Workflows, Not Another Shift
Trend of the Day|August 27, 2026
Taiwan’s retail and service industries are clearly shifting their focus from “opening more stores” to “improving labor productivity.” Recent reports on August 25 pointed out that convenience stores and supermarkets are adopting self-checkout, electronic shelf labels, and mobile payment systems to free employees from repetitive tasks such as checkout and price updates. At the same time, the ASUS EXPERTHUB service and retail forum held this afternoon on August 27 highlighted an even broader point: digital transformation is no longer about adopting a single tool. It is about connecting operations, scheduling, sales, membership, and management systems into one integrated workflow.
This trend is even more relevant when applied to the food-service industry.
In an era of labor shortages, true competitiveness is not necessarily about who owns the most machines. It is about who can turn an effective workflow into a repeatable model across more locations.
In the past, retail competition often focused on how fast a company could expand, how many stores it could open, and how much sales each location could generate. But by 2026, people’s time has become an increasingly valuable operating resource. The purpose of self-checkout, electronic shelf labels, and mobile payment is not to remove people entirely. It is to reduce the time employees spend on repetitive work, so they can focus on tasks that truly require judgment, interaction, and service.
Food service faces the same issue.
If a workplace, hotel, or campus still has 20 people looking for hot meals at 10 PM, the traditional response is usually to add one more staff member, extend kitchen hours, or set up another meal-service counter.
But when that situation expands from one location to five, ten, or twenty locations, continuing to support every site by adding more labor becomes increasingly difficult. Management cost, staffing coordination, and quality control all become harder to maintain.
This is the problem that 運吉 aims to solve:
Can we first turn the process of how a hot meal is stored, purchased, reheated, and collected into a repeatable workflow?
Let’s use a cheese scallion pancake as an example.
A cheese scallion pancake clearly shows why different foods should not be forced into the same heating method. The pancake needs a fragrant and lightly crisp outer layer, while the cheese and filling must be thoroughly heated. If the entire product relies only on microwave heating, the cheese may melt quickly, but the crust can lose its crisp texture as moisture redistributes.
If the whole product relies only on steam, the pancake may retain more moisture, but for a product that should have a toasted, crisp outer layer, too much steam may not be ideal.
That is why 運吉 does not set every product to the same “microwave for three minutes” program.
Instead, we test the actual cheese scallion pancake that will be sold, then design the reheating program according to the product’s real characteristics. For example, microwave energy can first help heat the cheese, filling, and center, while an appropriate infrared heating stage can be used to improve the surface texture and bring back a more toasted, crisp finish.
If this product does not need steam, the steam stage does not have to be activated.
If the next product is a steamed meat bun, it may require microwave plus steam. If it is a soup product, the heating sequence for liquid and solid ingredients may need to be arranged differently. If it is rice, risotto, or noodles, another set of parameters should be established.
This is the true value of hybrid reheating:
More heat sources are not automatically better. Each food should use only the process it actually needs.
The storage strategy should also be flexible.
Depending on the product’s storage validation, packaging method, delivery distance, and actual sales turnover, storage can be planned from frozen conditions to approximately -8°C to -10°C low-temperature or mild-frozen conditions.
For example, products with high lunch demand can have higher replenishment frequency and faster turnover. Meals prepared for evening or night-shift use can be stocked in smaller quantities. One office building may need 20 portions per day, while another may need only eight. They do not need to follow the same inventory logic.
Therefore, in the first stage, 運吉 does not require a company to implement a full and complex AI platform immediately.
Start with the standard system.
The central kitchen or food brand first handles product preparation, portioning, packaging, and low-temperature delivery. The 運吉 system then handles on-site storage, self-service selection, payment, meal-specific reheating, and pickup.
The first goal is to answer the most important operational questions:
How many portions are actually sold each day?
At what time do people buy?
Which turns over faster, cheese scallion pancakes or rice meals?
How many portions should each site stock?
How often should replenishment happen?
After the first and second machines begin actual operation, additional functions can be added as needed: AIoT, temperature records, inventory management, sales statistics, abnormal-condition alerts, replenishment reminders, and remote management.
At that point, managers are no longer just checking whether the machine is turned on. They can gradually see more valuable operating data: Building A in the Tainan Science Park sells an average of 12 portions after 8 PM; Building B has lower demand on Fridays; one machine has only four portions remaining; one product has increased in sales for three consecutive days; one machine has an abnormal temperature condition that requires inspection.
When these data flow back to the central kitchen, the next production and delivery plan can move closer to real demand.
This is exactly where service and retail digitalization is heading:
The goal is not unmanned operation for the sake of being unmanned. The goal is to stop people from repeating the same low-value tasks over and over again.
運吉 can begin with one real deployment in hotels, technology companies, campuses, central kitchens, food brands, or startup product testing.
First, let one cheese scallion pancake product run successfully.
Then add a second and third product.
As more locations are added, inventory, sales, and AIoT data can be connected.
A brand does not always need to open more stores.
Replicating an effective meal-service workflow can also help the service go further.