Cognitive Automation (CA) systems are built using the power of an Artificial Intelligence (AI) platform, which offers a rich set of cognitive services [i] . Microsoft Azure AI, Google Cloud AI, IBM Watson and Wipro HOLMESä are some of the top competitive AI platforms in the marketplace [ii]. Business and IT/Infrastructure applications adopting CA, harness the cognitive services of the AI platform as relevant for their use cases. For accelerated development of cognitive applications, many of these AI platforms have abstracted their cognitive services as Application Programming Interfaces (APIs) and published them for quick integration with the customer ecosystem, while continuing to own the platform intellectual property (IP) rights. This API framework has simplified the pricing into an API-as-a-Service (AaaS) model where customers can be charged based on their usage (every invocation of the APIs) along with a licensing fee for using the IP. I shall further elaborate on this through a case study from a real customer data and illustrate the attractiveness of the AaaS pricing model.
API-as-a-Service (AaaS) – Price per API call
A very large electric company U embarked on a Configure Price and Quote (CPQ) transformation program to improve the quote productivity, win rate, and top line (revenue) and bottom line (profit) growth. A key milestone in this journey was to achieve differentiated level of business performance by embedding run-time actionable cognitive intelligence within the application to empower price managers (PMs) to improve decision making across the "opportunity to order" process by generating and delivering on-demand, actionable, context-rich, data-driven insights. Specifically, the cognitive solution provider V had to build a CA system to enable PMs to search for the most relevant comparable quotes for a new quote that needs pricing decision, using a large set of strategically priced historic quotes. Every search for comparable quotes run by the PM will incur an expense and the cumulative expense over the year will be invoiced to U. The following describes a mathematical approach for this AaaS pricing model. Let’s begin by defining the variables and equations to compute the price per API call and the yearly expense for U (also the revenue for V) realized from the total volume of API calls invoked in that year.