API Series – SAS: APIs and the Rise of the Analytics Economy
This is a contribution for the Computer Weekly Developer Network written by Paul JonesHead of UK&I Technology at SAS.
SAS is known for its work in data analytics, data management and artificial intelligence (AI).
Jones is a regular speaker at industry events and a published author.
His current role is to help organizations address their data and AI challenges by adopting a transformative, enterprise-wide analytics strategy to unlock value from their data.
Jones writes as follows…
APIs have fundamentally changed the landscape of analytics, making it quickly consumable. They are central to how we can now quickly deliver analytical business value in the cloud. [and many other services besides] using lightweight composable architecture principles, which allows us to seamlessly integrate new functions such as analytics into existing business processes.
Analytical APIs and their management are an ever-evolving art.
JThe more we learn and progress, the more we can benefit from various capabilities throughout the analytics lifecycle. This includes everything from collecting data from new sources to modeling and communicating the results to consumers.
Open big data, often powered by APIs, is now ubiquitous. Therefore, being able to extract data from multiple sources is a key need for new analytics models. Some of the most valuable information is obtained by extracting information from new or different data sources.
APIs are essential for this and as such analytical economics evolved rapidly with the data Economics of APIs.
All of this means that the role of data scientists has changed alongside it.
The most visible and significant impact of analytics-focused APIs is on citizen data scientists or business users who may perform some analytics themselves. Analytic APIs have affected this group in two ways: they provide open access to more data, but more importantly, they make it easier to deploy advanced analytics to affect business processes.
Enabling business users to influence a business process by streaming their AI through an API adds agility and impact, driving value, efficiency, and automation to those processes. It’s a real business game changer.
Democratize data and decisions
APIs have allowed us to place less and less responsibility for deploying AI models on IT teams. Self-documenting open APIs can be integrated into analytics platforms for publishing in an operational framework, allowing end users to automatically “publish” activities.
Lightweight Open Container Initiative-compliant Docker containers are created to execute AI decisions that are published in a container registry. These are fully portable, API-centric and allow businesses to run this IP anywhere in the case of a global enterprise with light infrastructure requirements.
We can now easily launch and maintain multiple AI models and do it at scale.
APIs in the future
There is still room for improvement and we are all on a journey of discovery.
Today, integration with enterprise-wide data sources is still relatively limited, but I predict that over time this will change.
We will continue to see even more efficiency and simplicity in deployment models as real-time interaction becomes increasingly sophisticated.
API deployment can also remain a reasonably technical process, especially the last step. Normally this requires a set of IT skills to deploy the container image. There is an obvious possibility to make this even easier. We’re also working on the metadata that sits in the AI and APIs to make sure we’re on the cutting edge of accountability and compliance.
There is no doubt that AI-related APIs are realizing their enormous potential in changing the way we collect, manage and use data. We have seen more and more citizen data scientists in enterprises (and those just exploring) and the evolution of APIs is what has made this possible.
SAS Via, which is cloud-native, is built on open APIs in many layers. But really, it’s what our customers build using these assets and what can be deployed that’s important. Analysts talk about two-thirds of AI models that will never come online because IT teams don’t know how to deploy them. All of that is changing, thanks to easy-to-use APIs.
Over time, APIs will continue to evolve to enable even more real-time value – there is still room for growth and the AI applications this will provide could be staggering.