Building Smarter Apps with Cognitive APIs
May 2017 • Presentation
This session introduces cognitive APIs, pre-trained APIs, and trainable-models-as-a-service and showcases using cognitive APIs to build a conversational chatbot.
Software Engineering Institute
This presentation was created for a conference series or symposium and does not necessarily reflect the positions and views of the Software Engineering Institute.
This talk introduces "cognitive" APIs, such as those for image recognition, text analysis, recommendations, and predictions. I will explain what cognitive APIs are and how to use them and do a live comparison of four image-recognition endpoints. I will show the similarities in usage patterns and highlight the differences in API design, implementation, and capabilities, including pre-trained APIs and trainable-models-as-a-service. The second half of this talk is a live demo of an e-commerce chatbot that uses a combination of natural-language understanding and API orchestration to deliver commerce features over a conversational interface. I will then explain how the bot is built and where the "smarts" come from. I will discuss the ever-increasing importance of cognitive APIs and how no-screen interfaces like chatbots will stimulate the API economy.