Our demo site is now available. We used the Datacrag building blocks for assembling an assortment planning application which uses the Watsons retail data made publicly available during the A.S. Watson Hackathon of January 27-28, 2018.
To get access, please send us an email at email@example.com and we will create a user account for you.
The Datacrag application implements the concept of Augmented Intelligence, which combines the benefits of human experience, collective knowledge from experts and artificial intelligence.
We believe that AI has the most value by acting as an assistant to the business. In this sense, we designed the application to enhance human intelligence rather than replace it.
The key is to position the AI components at the right place in a given business process. Done correctly, the business gains become significant in terms of cost savings, improved workflow efficiency and decision accuracy.
Our building blocks can be assembled in many different ways to match a specific business process.
We designed the architecture of the application to allow a rapid deployment. The assortment planning demo site for example took only 4 hours to assemble and publish on the our cloud server.
The following case studies show how we implemented the concept of Augmented Intelligence in practice.
Product Image Search
Imagine you are a merchandise planner working at a sourcing company whose main function is to allocate products received from brand customers to the right vendors. The traditional process is to look at the product attributes and try to match these with attributes of products sourced in the past. This will shortlist the possible manufacturing resources (say from 500 vendor options down to 25), however there is still an additional manual step during which you would need to look at the products these 25 options have manufactured in the past and try to identify the most appropriate one.
If you are working with thousands of vendors and have a large database of products, it is very difficult to make a good sourcing decision by considering all the available factors. The process is obviously time consuming as well as expensive.
This case study is a perfect example on how Augmented Intelligence generates direct value to the business.
The building blocks needed for covering this particular business process are as follow:
- 13 workflow blocks to integrate and transform the source data from PDF and XLSX files
- 19 UX blocks to build the interface
- 1 AI block
The application uses a drag & drop file uploader to allow users to upload any file to contains an image of a product and compare it against all product images that are in the database. The AI block, which uses a siamese network in the backend, will then rank the images according to physical similarities and return a searchable table containing the 100 most relevant products with their attributes.
As a result, you as a merchandise planner can now review the table and make the allocation decision by prioritizing vendors according to country of production, FOB, capacity or manufacturing knowledge required for producing the past similar products.