Foresight about data explosion led to the birth of this analytics startup
Around a decade ago, Vivek Vipul was piqued by a report which stated that 80% of data generated at the time was unstructured and unusable. He gauged that the problem of unusable data would become even more pronounced with the proliferation of new-age devices along with the increased use of technology across different business sectors and verticals.
It was this foresight that prompted Vipul to launch a startup called Abzooba Inc along with this Indian Institute of Technology-Kanpur batchmate Rajiv Pratap.
Founded in 2010, US-headquartered Abzooba offers natural language processing-based artificial intelligence solutions to enterprises for connectivity, storage, data-engineering and analytics needs.
“Abzooba has a systematic approach towards enterprise AI solutions using a framework called xpresso.ai,” Vipul, who is also Abzooba’s chief executive officer, told TechCircle. “This enables Abzooba to deliver AI projects using a reproducible methodology in a timely and robust manner. The enterprise AI journey starts with data intake to data preparation and cognitive modelling that leads to actionable insights.”
xpresso.ai has accelerators for each phase of the digital implementation (enterprise AI) journey that are built for specific cognitive goals. Depending on the cognitive maturity of an enterprise, Abzooba leverages the appropriate xpresso.ai accelerators to manage AI development and production deployment.
After the bootstrapped company was incorporated, half-a-dozen employees started developing the NLP engine and xpresso.ai platform. Co-founder Pratap went about setting up the company’s India subsidiary, Abzooba India Infotech Pvt. Ltd, to develop products and solutions.
“We took nearly three years to build the complete platform and after that we started increasing our strength in both countries gradually,” said Vipul, who had earlier founded a startup called BrachySolutions Inc which sought to enable cross-border cancer treatments.
Abzooba got its big break after pipping IBM Watson, the technology giant’s AI platform, to sign a deal with the Singapore government.
Vipul said the road to prominence was not easy and it took months before Abzooba signed up its first client in the form of US-based IT services firm UST Global.
UST Global also became the company’s first institutional investor, helping Abzooba raise $6 million in funding.
“I had to make close to 65 pitches to investors in the initial days of the company. Our working capital was very low. We even came close to delaying salaries but the funding came through in time and since then we have had no such issues,” Vipul said.
Vipul said that he and Pratap had started a services division owing to lack of investments early on while continuing to add features to Xpresso.ai.
“Our revenue model is services-based and customised for clients,” Vipul said.
Abzooba has more than 20 clients spread across industry verticals such as retail, healthcare, banking, financial services & insurance (BFSI), and fast-moving consumer goods (FMCG).
Around 95% of the company’s clients hail from the US. While India’s contribution to revenue is negligible, around 65% of Abzooba’s 250 employees work out of its development centres in Pune and Bengaluru. The Indian unit reported a net profit of Rs 96 lakh on total income of Rs 17.45 crore in the financial year 2017-18.
The company competes with the likes of Mu Sigma, Fractal Analytics, ITC Infotech and Accenture.
“There is significant competition as well as collaboration in this segment which is worth more than $200 billion as an addressable market size. Plus, the addition of analytics is expected to drive up the figure significantly,” Vipul said.
According to a ResearchAndMarkets.com report, the big data analytics market was worth $8.5 billion in 2017, and is expected to grow at a compounded annual growth rate of 29.7% to $40.6 billion by 2023.
In terms of revenue, Abzooba is reliant on its platform xpresso.ai that accelerates development of applications in the professional services segment.
Xpresso.ai mainly consists of six accelerators put forth as a base for clients – xpresso data connectivity, xpresso data storage, xpresso data engineering, xpresso cognitive framework, xpresso solution components and xpresso infrastructure.
Vipul said the data connectivity accelerator connects any data source available in structured, un-structured and streaming formats, enabling easy creation of custom connectors for any APIs, databases or file-based formats.
As for the data engineering accelerator, it uses an architecture capable of consuming various data sources in a fast and inexpensive manner.
“Multiple internal and external data feeds within enterprises from various sources can be processed in parallel and merge a wide variety of data coming in at high velocity and high volume,” Vipul said.
Speaking about the data storage accelerator, Vipul said that it can be used to funnel all data sources into a storage layer after systematised validation and cleansing.
“The storage landscape with different storage types and extreme flexibility is built-in to manipulate, filter, select, and co-relate different data formats,” he said.
The infrastructure accelerator offers the ability to process large volumes of data parallelly. The cognitive framework accelerator leverages deep learning and machine learning libraries to build a flexible cognitive framework with the ability to integrate external packages through API and flexibility to handle automated feature engineering, model parameter optimisation and get actionable visual business insights.
The solution components accelerator, according to Vipul, works on a microservices model and can be used for solutions such as natural language understanding, semantic knowledge engine, recommendation engine, information retrieval engine, and summarisation engine, among others.