We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – August 3. Join AI and data leaders for insightful talks and exciting networking opportunities. Learn More
Unlocking data trapped in millions of documents – including PDFs, images, handwritten notes, audio, videos and emails – is hardly a new challenge for organizations. To meet that challenge, Indico Data is using artificial intelligence (AI)-powered deep learning to extract data stored in unstructured formats and making it meaningful.
Unstructured data is one of the most untapped sources of valuable information. In fact, MIT reports that 80-90% of all company data is unstructured. Unlocking even a small percentage of that could give companies a competitive advantage, experts say.
Adding AI to unstructured data
Indico Data’s AI-powered unstructured data platform is designed to help enterprises extract and make sense of the data in formats that otherwise would be next to impossible to define. The company today announced the launch of a new tool, the Indico 5.
“What makes us different is the core technology that we’ve built around what we describe as ‘composite AI,’ Indico Data’s CEO Tom Wilde told VentureBeat. “The underpinnings of composite AI are essentially one part deep learning, one part data, one part algorithmic all synthesized together combined with the application interface that we built.”
Indico Data claims to be unique in that it has an overall 95% success rate in deploying its AI technology and getting it to production — in contrast to the industry, which sees about an 80% failure rate when it comes to deploying AI, the company says.
The business-to-business aim of Indico Data’s platform is to close what the company refers to as the “impact gap” that lies between the unstructured data in various formats and give customers the tools to make meaningful insights from that data in comprehensive, efficient ways and actionable ways.
“In five years we’ll look back and it’ll be obvious that unstructured data had to be solved as a problem unto itself because it’s so difficult technically,” Wilde said. “At that point I think nearly every major Fortune 1000 company will have an unstructured data strategy in the same way that they have a data warehouse and a data lake strategy and an analytics strategy now.”
Bringing structure to the unstructured
It takes only about 200 documents to train an Indico model to extract data with about ultra-high accuracy, according to Indico Data’s website, “That’s because the Indico platform is built on a database of some 500 million labeled data points, enough to enable it to understand almost any document or image.”
The technology then uses AI, ML and natural language processing — in more than 70 languages — to translate the unstructured data into what’s needed for a specific customer’s use-case.
Indico 5 boasts several new features that promise to make this experience more seamless for its customer base including automatically linking and making labeling connections; splitting and unbundling even the most complex documents; universal document support that recognizes handwriting as well as text; and a workflow base that allows users to review each step of the automation process and tailor it to their needs with a low-code foundation.
Among other features, Indico 5 trains on a staggered loop, which allows the AI to review any corrections made by humans in the process and learn from that information for improvement next time.
The ‘unstructured data imperative’
Deciding to take on unstructured data was by no means a light decision for the Indico Data team. The company claims it has nailed down the complexities of disrupting the unstructured data market and its plans for growth don’t stop there.
“We call it the unstructured data imperative. For decades, people haven’t tackled unstructured data as a strategic imperative because they couldn’t,” Wilde said. “Trapped inside that opaque data is real value in terms of creating new products, improving the cost efficiency of the business and making better decisions from this new set of data. That creates more insight into specific activities like insurance underwriting, for example. It’s all about data in and data out and that’s how you build a risk model. And that is an example of what suddenly unlocking this new set of data can improve.”
Indico Data has its sights set on scaling further and expanding its industry reach from solely insurance, finance and real estate to include healthcare and life sciences — both of which have vast amounts of unstructured data waiting to be tapped.
