We understand the need for a data oriented approach to improving the core algorithms in your applications using AI/ML. We can help you harness the value of your business data and infusing it with AI algorithms for a smarter and intelligent business-specific solution that can help you to get more proficient in reacting to your customer needs.
Adding machine learning to your business processes increases productivity, business process automation, and more accurate business outcomes. We are experienced in use of modern machine learning tools such as Tensorflow, Pytorch and the AWS ML stack. We are also experienced in the data preparation for model generation and for continuous model improvement.
We recently developed a cloud based customer loyalty application that used an AI based approach for a customer to quickly to add the loyalty points whenever a user bough a beverage or a snack. The technical challenge was to scan the unique code on the drink can and the snack bag: we developed the AI solution that used the mobile device camera to capture the unqiue label from the can or the bag.rough an app. What made this project particularly challenging was that users would scan screenshots from all angles, different lighting conditions, different image qualities. Moreover, for the snacks, the bag would be crumpled as well which added to the difficultly. Our solutions created a 98% accuracy for label detection.
In this case the problem was to build a product which was fully automating the user acceptance testing using AI Bots that acted like humans. The bots were trained to navigate the app autonomously, run the various use cases on each page and then report any problems that were encountered. This would make it easy for the developers to reproduce and fix problems. Manual automation of these tasks can take 6-8 hours to automate a user journey but with this approach the problem can be done in minutes. The AI Bots automatically interacts with your app to discover operable flows and then executes the use cases on each page by emulating screen readers and using the keyboards. No need to manually go through screens. No need to write scripts to automate user journeys. To implement this product we developed a computer vision based model for object and text detection and screen classification.