Telecom analytics solution provider Subex recently launched an augmented analytics platform — HyperSense, to change the way companies access and explore their data. Analytics India Magazine spoke to Rohit Maheshwari, Head – Strategy and Products at Subex, to understand how this recent innovation is democratising AI and driving data analytics adoption.
“Using AI in business is not straightforward. There are issues around having data which is organised in a way that AI can leverage,” said Maheshwari. “Empirically speaking, around 70 to 80% enterprises struggle with uniform AI and data stack, along with attracting and retaining great talent.”
To secure data in the cloud and make it AI-ready for businesses, Subex has also partnered with Snowflake.
Excerpts:
AIM: How is augmented analytics encouraging citizen data science?
Rohit Maheshwari: The root of augmented analytics lies where businesses start solving some of the most challenging problems in the data, whether it is discovering data, preparing data, leveraging concepts of AutoML, leveraging AI to generate insights, or asking the most difficult questions from data in natural language, and then getting responses in natural language. The critical question is, can we make data speak to businesses leveraging augmented analytics?
Secondly, while data scientists can write some of the most beautiful and sophisticated algorithms to solve business problems, data scientists need business critical thinking and need to understand the context of data and the cultural nuances of the business itself, and which data scientists are often far removed from; they love their algorithms. In such scenarios, augmented analytics can bring data science closer to business. If augmented analytics can make business analysts and business critical thinkers do their data science, it can perhaps address this gap. This is why augmented analytics is becoming powerful.
AIM: Tell us about HyperSense. How is it democratising data science?
Rohit Maheshwari: While dealing with some of the seemingly impossible problems to solve in enterprises, we got inspired to build HyperSense. As a matter of fact, the desire to address this issue of solving complex industry problems triggered the desire to build this augmented analytics platform. Thus, all the issues that I mentioned in my previous answer is what HyperSense is designed to do.
Further, one core reality of enterprises is that not all data can be moved to the cloud. There are several regulatory challenges in many scenarios, restricting businesses from putting their data into the public cloud. So the ability of businesses to access the existing cloud stack is limited in a way. Also, while businesses address some of the complex problems by transforming to the cloud, they can’t throw away all the plumbing and their systems overnight. In such scenarios, Subex’s HyperSense is designed as a cloud-native augmented analytics platform available to enterprise either on-premise, in hybrid situations or the public cloud.
Alongside, our 25 years of experience dealing with analytics, especially in the telecom industry, has taught us how to integrate and manage data at scale and solve some of the most complex problems. We are bringing all that learning to this augmented analytics platform for our enterprise customers.
Drawing an analogy — when we see a box of Legos, we see a car or a house already built in the front of the box to inspire the kids to build their own thing. Similarly, in HyperSense, we have several pre-built data operators, pipelines, applications and use cases to drive adoption and encourage business users to start using them automatically. Thus, we believe all of this will significantly democratise data science and help organisations trust their data and decision making — creating an environment of digital trust.
AIM: What drove Subex to create an end-to-end augmented analytics platform?
Rohit Maheshwari: Today, if you were to try and assemble your own AI models, you get two groups.One group hit the open-source world. But, curating a working stack from the open-source world is a massive undertaking. Considering open-source is a very vibrant environment, it usually has ten solutions for every problem, making it difficult for data scientists to navigate and curate the right projects. The second group looks at the data analytics chain, starting from data ops, data wrangling, traditional analytics, ML ops, AI-driven analytics to visualisation and workflow consumption, and assembles a range of products for the model work. Either of the cases becomes a challenge for data scientists.
So the idea behind end-to-end is to address all aspects of data analytics and take away the pain of curating, assembling, and stitching these components together to build the AI model. At the same time, to provide more flexibility to our customers, we have made it composable modular. This allows the customers to either use some components of HyperSense in a subscription environment and for other components, they can look at other analytics platforms.
Does that help? The answer is — yes! Yes, it does.
AIM: How does Hypersense help enterprises tackle AI/ML challenges?
Rohit Maheshwari: As we have discussed before, solving business problems would require business critical thinking and an understanding of the business environment in which the problem exists. It also requires an understanding of data as well as the cultural nuances. Thus, data science alone can hardly solve a problem. With HyperSense, we wanted to enable business critical thinkers to solve problems independently and thus democratise AI.
HyperSense can solve problems, enabling a drag-and-drop no-code environment, where people can build their own pipelines and solve the problems. Further, HyperSense also comes with pre-built assets, such as pipelines, example applications, use cases, and ML engines, designed to be directly consumed. These can also act as an inspiration to build their own codes. This alone would enable enterprises to overcome some of the skill challenges.
AIM: What makes Subex’s augmented analytics platform stand out?
Rohit Maheshwari: While HyperSense has been built to play in multiple industries, we realise the importance of massive data in the telecom industry. Thus our port of call has always been telecom companies. Alongside, in the current era, telecom business covers multiple businesses — from near connectivity and content to telecom companies transforming into fintech and ecommerce companies.We believe that telecom companies are now becoming platform players, which allows us to bring different industry use cases in one place. Thus HyperSense can be differentiated in terms of scale, volume and the ability to solve multi-disciplinary problems. Further, HyperSense comes with several pre-built use cases, including 5G, edge analytics and IoT, which also helps us stand out against competitors.
AIM: Tell us about the core tech stack
Rohit Maheshwari: HyperSense has been built using a curated open-source stack that is not only cloud-native but also Kubernetes-native and containerised. Also, we are using a number of open-source stacks which are scalable, such as S3-Compatible Object Storage for software storing.
Further, for the front end, our developers used an open-source web application framework — Angular, for us to stitch components and scale to emerging requirements. So, if a superior technology arrives, we will be equipped to quickly adopt new technology without impacting the customers. Also, the entire thing is built on a microservices-based design, using CI/CD to continuously provide updates and upgrades to our customers.
AIM: What are some exciting use cases for HyperSense?
Rohit Maheshwari: HyperSense can solve many interesting problems from a variety of industries —
Telecom industry: HyperSense can solve use cases around sales and distribution, marketing and network operations, support systems — such as human resources; finance; fraud; business assurance; risk management.
Manufacturing industry: HyperSense can solve use cases around sales and marketing; knowing where to run campaigns; precision marketing; forecasting demand; revenue analytics; predictive maintenance; inventory management; supply chain and logistics; quality management; and research development.
Insurance industry: HyperSense can solve use cases around risk management and fraud management; pricing and product analytics; claims analytics; churn, and customer retention analytics.
