All articles

Breaking the Data Barrier to Digital Transformation

My first job taught me a lot about technology and disruption. I worked for an investment bank as a financial consultant, and I watched how online trading platforms changed the investor landscape by simply offering a better customer experience. Technology shattered the monopoly that legacy brokerages had on information and access to make trades. Investors responded by taking their portfolios into their own hands.

The ability to trade online for pennies per share was a significant incentive, but I saw there were bigger issues at play. The large venerable brokerage firms offered a high touch model, with individual financial consultants building personal relationships with investors to sell them standardized products. That was the exact opposite of what their customers wanted.

Investors wanted a self-service model that allowed them to buy personalized investment products, so they embraced online trading. Many of these venerable brokerage firms, stuck with a 20th century business model, did not survive the financial crisis and were bought for pennies on the dollar in 2008.

The financial crisis was a big shock, but the same trends are playing out in a wide variety of industries. Pierre Nanterme, CEO of Accenture, laid down the gauntlet while speaking at the 2016 World Economic Forum. He said, “Digital is the main reason just over half of the companies on the Fortune 500 have disappeared since the year 2000.”

I embraced technology sales and began working to help businesses in all sectors transform themselves to become more agile, digital and customer-centric. I worked for a Robotic Process Automation (RPA) company and learned about the immense promise of digital labor and its potential to change how companies execute work. I also saw the limitations of RPA. These rules-based macros could only work with data inputs that were formatted for machines to read, but most documents are designed for people to read. And existing technologies like OCR and NLP were not flexible enough to fully comprehend these documents and extract the valuable information contained within. That simple problem has been a significant hurdle for businesses trying to optimize their operations and compete as a 21st century company.

Several studies from analysts like Deloitte and KPMG show that most enterprises have adopted automation technology, but less than 20% of those have achieved scale beyond a few tactical automations. There are several contributing factors, but chief among them is that RPA requires structured inputs, while most enterprise data is unstructured. 60%-75% of the data most companies is un-useable, stuck in financial statements, emails, log files, customer call records, notes, presentations, old documents, etc.

The International Digital Corporation predicted that organizations that could analyze all relevant data and deliver actionable information could achieve an extra $430 billion in productivity gains over their peers by the end of 2020. These organizations would have the opportunity to take information that was previously hidden or unknown and turn it into powerful insights, leading to new opportunities, reduced risk, and increased return-on-investment (ROI).

The less said about the year 2020 the better, but the good news is that technology marches on and Moore’s law still holds true. Advances in compute power and neural network architectures have brought us to a moment in document cognition similar to the mobile device market in 2007.

In 2007, Blackberry was the most valuable company on the Toronto Stock Exchange, trading as high as $150 as subscribers exceeded 10 million. Nokia reached a peak stock price of $40 per share in 2007, selling 250 million of its Nokia 1100 phones.

In its 2006 fiscal year, Apple had $19.3 billion in total revenue. Sales of the iPod represented 50% of that number and sales of the Mac represented 38%.

In June 2007, Apple released the iPhone. The original iPhone sold just over 6 million units in its first year.

In July 2021, Apple reported $39.57 billion in revenue from iPhone sales alone for its fiscal Q3. One quarter of iPhone revenues was twice as much as Apple’s total annual 2006 revenue.

In early 2007, the conventional wisdom would have suggested that the mobile phone market would continue to be dominated by companies like Blackberry, Nokia, and Motorola. But Apple turned that conventional wisdom on its head.

In a similar vein, Applica has pushed Natural Language Processing forward with a proprietary neural network architecture called TILT, which simultaneously learns layout information, visual features, and textual semantics. This is incredibly important, because business documents often have formatting that requires NLP, Computer Vision, and Layout Analysis working in concert to understand them. Humans can see the relationships between text and layouts without issue, but NLP solutions that rely on text sequencing cannot preserve the relationships between extracted values when they export their results. Thus, these solutions cannot handle complex documents with nested tables.

In addition to this groundbreaking layout awareness, Applica has transcended the old paradigm that requires hundreds or thousands of manually annotated samples to fine tune from open-source backbone models for specific use cases. Users interact with our language model via text interactions to tell it what information you want from the document. You ask it questions and it provides answers.

Applica requires much less data for training and much less up-front effort. That means that customers can, at last, address document challenges in their organizations at scale, with as much as 90% less effort.

Better models yield better results with less effort. And just like flip phones and early smart phones paved the way for the massive success of the iPhone, the last ten years of document digitization have paved the way for Applica’s two-dimensional, layout aware document cognition solution to help enterprises solve their most difficult transformation challenges.

My first experience in the business world taught me that customers want self-service options to buy bespoke products and services. To meet these demands, companies must be able to do two things well: 1) mine their data to understand their customer; and 2) execute transactions at machine speed. Since so much data is contained within documents, Applica is essential to achieving true digital transformation.

I am excited to join Applica because it has differentiated, proprietary technology that represents a major step forward in the state of the art. Applica has broken the data barrier to digital transformation and I am proud to be a part of it.

If you would like to learn more, you can reach me at