Unilever, a globally-renowned FMCG corporation faced significant challenges with its paper-based third-party logistics (3PL) processes, resulting in inconsistencies and inefficient procedures. Their full stocktake was time-consuming, causing operational inefficiencies and prolonged warehouse downtimes. To tackle these issues, they adopted the Powerhouse AI application to automate their cycle count process. The application enabled a fully paperless operation, streamlined inventory counts, and allowed automatic cross-checking with their ERP system. An experiment showed that the new AI-driven process was 29% faster and achieved 100% accuracy, significantly improving the corporation's logistics operations.
The customer faced a significant challenge in gaining more control over their third-party logistics (3PL) processes. The company's previous methods were deeply rooted in a paper-based workflow that led to inconsistencies and immature procedures. This outdated mode of operation was not only manual but also time-consuming. A full stocktake would take several days to complete, leading to prolonged warehouse downtimes, preventing goods from getting in or out of the warehouse. The combination of these factors resulted in operational inefficiencies that the customer needed to address.
To address the operational challenges, we configured the Powerhouse AI application to support the company's cycle count process.
This allowed the company to transition into a fully paperless mode of operation, using image capture for inventory counts and automatic cross-checking with their ERP system, a change that significantly improved the productivity of their warehouse operators.
The solution was designed around a user-friendly process guiding the user from location to location. Counting and checking was enabled by taking three quick steps: (1) the operator would initially capture an image of the location tag, (2) followed by a photo of the label (note that most labels were not barcoded, requiring the application to extract key data points like SKU and lot number based on text). (3) The final step involved taking a picture of the pallet. Subsequently, the application autonomously cross-checks with the system data.
The app had to operate within the constraints of their storage setup, which permitted only a one-sided view of the in-rack pallets. In addition, it had to manage pallet wrapping and the absence of barcoded labels for goods identification.
The application was further tailored to record defects such as damage, mold, or leakage of goods, demonstrating its potential for broad operational use and offering a significant step toward automating and optimizing their logistics processes.
In an effort to assess the effectiveness of the new solution, the customer implemented a controlled experiment. This involved a random selection of 56 pallets, which were then counted and checked for both speed and accuracy. The same procedure was then replicated using Powerhouse AI, including all aspects of the process from counting to administrative reconciliation.
The results were compelling. In terms of speed, using Powerhouse AI proved to be 29% faster than the traditional manual method. The complete checking of 56 pallets, inclusive of driving time, was achieved in just 43 minutes.
Perhaps even more impressive was the improvement in accuracy. The automated process with Powerhouse AI yielded 100% accuracy, surpassing the 98% accuracy rate of the manual process. This combination of increased speed and enhanced accuracy demonstrates the significant impact the application had on streamlining and improving their operations.
