The developed platform replaces the manual catalog maintenance process, which currently has around 30.000 items, and reduces product registration time by 84%.
Benefits
- Catalog update completed in just a few clicks.
- Item registration time reduced from 13 minutes to 2 minutes.
- More detailed item description.
- Expected growth of 10% in online searches.
Overview
A Gimba Gimba is a leading Brazilian retail and supply chain company, responsible for the networks of 300 of the 500 largest companies in Brazil, scheduling all deliveries, quality standards, and defining SLAs for replenishment. With the support of AWS and Flexa Cloud—an AWS partner that uses Generative Artificial Intelligence (AI)—Gimba developed a platform to automate the steps involved in building its product catalog, which contains approximately 30.000 products.
About Gimba
A pioneer in the supply chain and distribution management market, Gimba helps businesses and consumers obtain essential items for their daily routines.
Opportunity
Gimba's CTO (Chief Technology Officer), Daniel Arruda, explains that catalog updates were previously done manually for all approximately 300 new products added monthly.
“Each manufacturer sends us their product data in different formats, and we have several collaborators who read and standardize this data, creating product descriptions in our catalog, especially the online one,” he explains.
João Ricardo Miliozzi David, marketing analyst at Gimba, also adds that "there is a concern with how the product is described to demonstrate Gimba's personality, inform the consumer about the main characteristics of the product in a unique way, and be more attractive to search algorithms, simultaneously."
According to Arruda, as long as other tools of Generative AI When they arrived on the market, the team responsible for updating the catalog began experimenting.
“We used them in a non-methodological way to produce some texts, which were then reviewed and improved by the team. This process was manual. The questions were not standardized, nor was the use of the answers,” he says.
With the goal of increasing productivity and improving the quality of product descriptions, Arruda's team saw AWS as an opportunity to use Generative Artificial Intelligence (AI) more efficiently and scalably. Furthermore, AWS would be able to take the next step, personalizing the AI according to Gimba's so-called "personality."
“We have deep confidence in AWS as a strategic partner in our first joint project.”
Daniel Arruda, CTO of Gimba
Why AWS?
After presenting the project to AWS, Gimba was invited to develop a platform tailored to its needs and bring the catalog solution to the AWS cloud.
With the support of Flexa CloudThe first prototype, based on Amazon SageMaker, was created for training and fine-tuning the LLM model. A sample of 900 products with ideal descriptions was used to train the model so that it understood what is expected.
“We tested two or three new products during the learning process and the results were impressive,” says Arruda.
“We knew we were on the right track with this first success,” says Deivid Bitti, CEO of Flexa Cloud.
"When we gained access to Amazon Bedrock, we began the transition to AWS managed services."
The chosen model, Claude-2, was fundamental to the platform's success due to its large context window (up to 100.000 tokens), which allowed us to use advanced out-of-the-box engineering techniques that eliminated the need to manually train or adjust the model. As a result, we reduced the cost of the solution by more than 50%.
The entire development process with AWS was based on the principle of customizing the use of generative AI to meet Gimba's needs, specifically those of the registration team. We created an online platform with a single, easy-to-use interface that automates the application of Amazon Bedrock APIs and immediate creation on the back-end.
Results
Now, with just a few clicks, the product catalog is updated within this interface, which automates a series of adjustments that were previously done manually.
“The platform already uses our communication standard and HTML markup. It's much faster; there's no comparison to the process we had before. We reduced the registration time from 13 minutes to 2 minutes per product,” explains Juliana de Freitas Ribeiro, registration manager.
On the other hand, Gimba's customers also gained access to a more complete and informative description, eliminating questions about the product and increasing sales conversion.
“Internally, we expect greater productivity in generating registrations and better-positioned organic searches based on the improved use of keywords,” reveals Daniel Arruda, citing an expected growth of 10% in these searches.
Next stages
With the success of the new product platform, the registration team expects to move all products currently in the catalog through this tool and revamp existing listings.
“By doing this with new products, we free up the registration team for other tasks, because the description takes more time. In this way, we are able to raise the quality standard.”
Juliana de Freitas Ribeiro, Registration Manager
"We have great confidence in AWS as a strategic partner in our first joint project."
Daniel Arruda, CTO of Gimba
