Just Analytics? Birlasoft says: No Thanks!

MUMBAI, INDIA: Not so once upon a time, there was a hay of data (structured, semi-structured, unstructured, butchered, glossily-wrapped, domesticated, chaotic – you name it) and there was this nail of ‘insight’ that was to be pulled out from a hopelessly-tangled knot of business information. Someone handed over a magnet called ‘analytics’ and Voila! The nail was easily, effortlessly slurped out. Things seemed happily-ever-after for some days.

Until the day when the mess of hay turned into an intimidating pile of nails and the challenge became utterly and impossibly formidable -To pull out one precise nail that fit one’s hammer from this eternal heap of iron.

Would painting some specific nails help? Would a different, smarter, meaner magnet come to rescue? Or would it have to be a case of bleeding one’s fingers all over again unless one was fine bleeding the business walls instead with endless holes and botched nails?

Dr. Amit Shekhar, Chief Strategy Officer, Birlasoft caught himself thinking something radically different and feeling restless with the magnet of analytics much before these questions popped. He was not only desperately hungry and imaginative but also unrelenting with the idea of pushing a new adjective before the noun analytics: ‘Predictive’.

‘Predictive’ is precisely how his team and the management too, dreamt analytics to turn out as. They just draped their ambitions in the fabric called ConnectedBI, and nailed the ultimate goal of ‘why can’t a business user magically pull out answer to a random question if we do it daily on Google?’

Dr. Amit Shekhar, Chief Strategy Officer, Birlasoft

Ask Srikanth Ramakrishnan, AVP, Business Solutions at Birlasoft and also the chief architect for the Connected BI solution, as to how much of this elusive rabbit out of data hats they have been able to catch and he beams. “Our advanced analytics solution – ConnectedBI Framework is a platform that can provide actionable and predictive insights on business performance by seamlessly mining data from structured and unstructured data source through its Semantic Knowledge Engine. The framework can track, observe, reflect and predict business performance and significantly increase the capabilities of existing BI investments and also allow automated actions triggered by insight. Equipped with such smart use of collective knowledge, organizations can identify business threats and opportunities which might have otherwise gone undiscovered.”

So may be they changed the wall itself or maybe Dr. Shekhar and his IT pianists devised a new hypnotic tune to woo the nails they wanted - well, let’s find out more from the Piper himself.

The enterprise world has suddenly moved to a recent buzzword Business Intelligence (BI) to a new genre of Analytics in just a few months. Is it just a nomenclature change or something substantial? What is your sense of analytics in the enterprise space as it stands today? And why chasing the ‘proactive’ flavor of Analytics?

Most of our customers are Fortune 50 or 100 ones and I can clearly see the big shift of combining technology and business that is happening there. We have solutions across geographies and businesses and our edge is driving the value chain brilliantly. We take pride in being at the most important points of the story that our customers are scripting – and being with the whole story is something that a rear-view mirror does not necessarily equip strongly for. BI was good but it was horizontal and a rear-facing tool, something that was not adequate at the pace with which industry has galloped to Big Data and Advanced Analytics. So we wanted to bring Predictive Analytics to the table and introduce innovations in Advanced Analytics which would drive real-time business decisions and importantly provide predictive view for various business dimensions

How does this contrast to the BI and data scenario you had earlier?

Over the years, enterprises have invested heavily in ‘computing power’, data warehouses and other data management technologies that remain largely disconnected. These enterprises are experiencing disorder due to incremental new sources and types of Data, rise of Data Sciences, ever changing business demands to embed analytical capabilities in traditional transactional systems. The key constraint which technology leaders are realizing is that traditional approach of data warehousing is inadequate to address today’s Data to Information needs. Corporations are looking for the ability to quickly index, search, analyze and act on information to drive business results. With the Connected-BI Platform we aim to assist these enterprises in unlocking the potential of their unmanageable data through efficient use of technology and increase the ROI of their existing data management investments.

Exactly how?

The platform provides customized data exploration of structured, unstructured, and semi-structured content, from multiple disparate databases or data residing in new Enterprise Data Lake (s)/ Big Data Systems. It offers capabilities in semantic and keyword search in real-time, querying large disparate datasets using minimal hardware and with minimal constant dependence on IT engineers. In short, business users can have conversations with their data - Ask Natural language questions, get instant answers.

Why would it be a right-timed idea for an organization like yours?Is the solution amenable to be tweaked for a specific micro-vertical if required since your group entails a big and diverse mix of such verticals?

Birlasoft, that is a CK Birla Group company (a $1.6 billion diversified conglomerate with a strong footprint in manufacturing, engineering, technology, hospitals and education), does have a strong micro-industry focus and we continue to strengthen solution capabilities for specialized solutions by varied industry segments. Birlasoft deploys a host of innovative solutions and service architectures across the customer value chain so yes, we have some interfaces already built in and some would be developed ahead this year. The solution is open-ended and POCs (Proof of Concept) show that it can work with a number of other technology boxes. Optimisation of visualization is currently underway.

How critical are algorithms in modeling scenarios like these?

The algorithm is the brainchild of any engine. One has to end up spending time on algorithms because the parameters and decision-criteria for every problem or domain would be different. That is where domain people come in. Partnerships with institutions like IIT or technology vendors who have engines built already, can also be considered. We also do the first cut of correlation and then we deep dive into next set of correlation questions. Yes, algorithms are the backbones of models.

Was it solely a DIY approach or did you pick some shelf-ware available in the market these days?

The lifecycle of any technology is generally two to five years. In some cases you build your own solutions and some cases you pick it from the market. We have used the best as it applies so at some places we have just used a wrapper around and somewhere we have used data parts or tools. To give you a cross-section view, the solution was developed using 100 per cent JAVA with Enterprise standards (J2EE, W3C, TOGAF), and is a BI Platform with extensible connectors and faceted Search, has SOA extensions for extensibility and scalability, has multi-tenant application support and is a statistical modeling framework by hosting R statistical runtime and with enterprise-level security.

Our solutions though, can talk back to other tools and interfaces and it is very much an open agnostic box. The system is based on Open Ended Architecture (not a Black-Box) and hence can easily be extended by in-house IT teams. The system can be hosted within Cloud- Applications such as Salesforce, hence, reducing the integration time between brick and cloud applications dramatically.

Where do you see its impact zone at Birlasoft?

We started a year back and put in place some POCs (Proof of Concept) in the middle of the year and have touched a wide scale of domains from Aviation, Automotive, Healthcare, Banking & Insurance to Oil and Gas. The impact and market feedback has been very positive. We have tested and delivered a range of abilities across semantics, statistical models, feeding all that into a strong engine. The supply chain will become very efficient with information-visibility at a new level altogether at various parts of the value chain we cover.

Can IT have a revenue-angle too?

Why not? If a forecasting tool can show insights from unstructured data on our end customers’ demands and needs than it can help optimize inventory cycle or reduce holding costs, impact top line, hence, affect revenue too.

What are the positive dimensions of this solution now if you were to slice it with a technology scalpel?

The best part is that Connected BI empowers business users to freely explore information and harness the power of combining all their data assets, including Big Data. This Data Mash-up technology allows data to be a shared asset among groups of people with differing roles and differing interests – who want to get different insights from that data.

So it can be built up upon easily and organically?

This is a unified, open platform and enables the creation of knowledge applications, solutions that combine the power of data with the knowledge of the business user. These knowledge applications can be easily customized based on the needs of the business users and shared across divisions, hence, bringing down divisional silos.

What makes it really distinct?

A key differentiator is Analytics which is based on Real-World Business Ontologies i.e. there is no hard-coded data schema. Business can directly define the ontology behind their Analytical needs; IT points the Platform towards where the data is dispersed in the enterprise. The platform then starts indexing and creating Analytical cubes and Data Marts based on the business ontology model. As business changes the indexes, analytical cubes/ marts automatically changes and hence enabling quick results. Further, the system has learning capabilities to understand the “why” behind search and tweak the data indexing and analytical process. This provides ability to deliver contextual answers to users' questions.

There is another breed of analytics making rounds these days – have you assessed Self-service analytics and its potential yet?

It is a relevant flavor of analytics as I see it. The self-service approach lets end users create personalized reports and analytical queries while freeing up IT staffers to focus on other tasks – potentially benefiting both groups. ConnectedBI solution has importantly achieved a critical intersection of Self Service BI and Semantic Technology for Natural Language Queries and helped cutting down BI Development costs by 60 per cent to 80 per cent. The cost reduction would differ from enterprise to enterprise and would depend on the reliance on IT engineers to support regular analytics.

That makes it all the more user-friendly?

Natural exploration of data is achieved by asking questions the way you would do in Google. This will empower you to freely get answers from your distributed data- structured or unstructured, which might be spread across your enterprise in relational databases, data warehouses, Big Data sources or in documents such as Excel, Word and PDF. The answers would be in form of reports, or analytical chats. For example you could ask “Show me the 10 most profitable customers between 2012 and 2014 and their respective Sentiment score”.

How relevant is the new trend of in-memory engines to advanced analytics?

I wouldn’t say the in-memory trend is completely new, but yes it is a little different and cheaper. HANA or connected BI is about intelligent scalability of the data itself. Data has a life and looking at it from in-memory, MDM point of view makes sense and makes the scenario scalable I guess.

What is looking exciting on Birlasoft’s map ahead?

Our micro-vertical strategy focus would continue. We are also working on a roadmap to better slice different pieces of a vertical’s value chain and integrate the flow across the entire chain. In general, the industry is warming up to new trends and I feel that IoT (Internet of Things) is here to stay. We are collecting a whole lot of data but we are now leveraging engines and correlation mechanisms too and that’s where we are ahead of the curve in the industry.