Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.
nRoad, a Boston-based AI startup that abstracts and incorporates unstructured data into vital business functions, has today launched a new platform known as Convus — aimed at helping enterprises tackle the challenges of unstructured data. Across the enterprise ecosystem, employees are building a bottomless data lake, premised on the corporate mantra to “save everything, just in case,” according to an article published in Gartner. Alan Dayley, a former research director at Gartner, notes that increased data growth over the past decade has created an unstructured data nightmare. “It’s not just the cost to store it. Huge volumes of dark data make it harder to find what is useful and may mean we miss business opportunities,” says Dayley.
Mike Gualtieri, VP and principal analyst at Forrester notes in an article that between 60% and 73% of all data within an enterprise goes unused for analytics. Gartner predicted that through 2021, more than 80% of organizations would fail to develop a consolidated data security policy across silos, leading to potential noncompliance, security breaches and financial liabilities. As organizations seek to build out strategies to address the growing unstructured data challenge, nRoad says its new platform is built and trained on long-form, highly unstructured documents and content. Designed to be form- and format-agnostic, Convus™ is powered by a set of modularized engines that are API-driven, says nRoad.
VentureBeat drilled into the intricacies of the platform via an email interview with nRoad’s CEO, Aashish Mehta, for a broader context on the platform and how it impacts the enterprise data infrastructure.
Eliminate the need to train and process unstructured data separately
According to Mehta, nRoad’s new platform eliminates the need to train and process various unstructured data separately, offering a singular interface (API) to perform any downstream tasks. This allows enterprise technology teams to break from widely available template-based solutions.
Mehta said Convus™ uses new vision-based and other deep learning algorithms combined with classical ML algorithms and an NLP layer that leverages publicly available transformer language models. The platform’s differentiating factor, Mehta claims, is the ability to incorporate domain knowledge using graph techniques and domain-specific language models that enables it to maintain domain context, relationality and semantic similarity of the data.
“The foundational flaw in generic approaches like RPA, OCR software, and cloud-based solutions to resolving the unstructured data problem is a lack of consideration in input variability, content variability, language, and localization constraints,” Mehta said.
He added that nRoad’s founding team has also developed a unique but standardized and intermediate representation of entire documents — thus enabling the reusability of the data and rapid deployment of downstream models for nascent use cases.
nRoad wants to function as a non-intrusive input source of unstructured data in a structured format, said Mehta. Convus™ is purpose-built to handle volume, velocity, variability and variety, he added. Built on the central philosophy to “extract once and normalize multiple times depending on the business use case,” nRoad believes enterprise CTOs and CDOs can leverage the domain-driven trainability capabilities of its platform to better contextualize data.
The company says its AI-powered Convus™platform is 2.8 seconds faster than popular cloud solutions, has a 90% extraction accuracy and, to date, has extracted over ten million domain-specific terms.
Unstructured data opens up a significant market opportunity
Mehta noted that several enterprises have invested huge resources on their business process management (BPM) and other tools to automate structured content over the last decade. However, even with this investment, unstructured content processing is still a major challenge and remains a notable frontier in their quest to achieve true digital transformation and hyperautomation. Given that the unstructured content represents the majority of the total content generated today, Mehta said it represents a significant opportunity for nRoad.
nRoad addresses the major challenge of automating unstructured content processing. “For example, if a major financial institution receives approximately 50 million documents a year, every time any unstructured content is encountered in the workflow process, the process becomes manual and prone to all kinds of inefficiencies,” said Mehta. “Our platform helps to process these documents automatically by extracting major variables and generating insights. These variables can then be fed to the workflow to ensure a frictionless process that creates no dark data.”
How nRoad differentiates itself from the industry
The three major types of players in the industry, according to Mehta, include:
- Cloud-based, horizontal platform providers who are focused on providing tools to enterprises to address unstructured content processing needs.
- Players focused on providing a document or a function-specific solution like mortgage documents.
- Business process outsourcing (BPO) companies focused on processing documents as part of their outsourced services, using tools and manual efforts.
He said nRoad’s differentiation is in its ability to provide a fully automated and scalable solution that can reside behind an enterprise’s firewall. He added that nRoad also empowers small and midsized enterprises (SMEs) to interact with the processed data to derive further intelligence.
While Mehta agreed that nRoad has competition in all three players, he said nRoad’s solution provides enterprises with a unique and low-risk opportunity to address unstructured content processing needs without losing control of their workflows, unlike anyone in the industry.
“Our platform can process super text-heavy documents to generate insights like private credits, ESG reports and PPMs, and this sets us apart from many players in the industry,” he said.
In its press release, nRoad claims its Convus platform currently serves the financial services industry, and has proven its capabilities to some of the largest fintech players, asset management firms, and financial data providers globally.
Upcoming milestones to expect
nRoad envisions a greater use of NLP techniques, graph techniques, machine comprehension and newer vision-based techniques to handle complex business use cases.
nRoad was founded in 2020 and has a current headcount of 35, but the company expects to double this number on the heels of its revenue growth and quest to accelerate the implementation of its product roadmap.
While Mehta did not comment on the exact number of customers nRoad services, he said the company grew by 175% in 2021 in terms of customer growth.
nRoad has been self-funded so far, but the company will initiate a funding round in the coming months, with claims that its clients and a few VC partners have expressed their interest to invest in the company.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.