To say 2020 has been a bumpy ride would be an understatement, especially with a pandemic that’s expected to cost the global economy over $1 trillion in 2020 alone. However, if there is a bright side, it’s definitely shining on the tech sector with an unprecedented global switch to online versions of pretty much everything. The document management software market, in particular, is receiving growing public interest and is expected to reach $996.8 million by 2026.
The sharp rise to its current position at $859.5 million this year is in no small part due to the massive migration toward remote workspaces. With more data coming in from more places and in more formats than ever before, organizations are under immense pressure to index, organize, categorize, and generally make sense of it all. This is where document management solutions come in and automate a lot of the repetitive drudgery involved with ingesting content and converting it to actionable insights. Here are four startups making noise in this hot sector.
While there are literally hundreds of document management solutions on the market, most still require a lot of human input, resulting in employees spending a considerable amount of time on data management and compliance. Putting artificial intelligence and machine learning to work in the field of document management is the first startup on our list, New York-based Hyper Labs. Established in 2014, Hyper Labs brings to us the world’s first “input-to-outcome” automated platform for document management called Hyperscience IDP.
The IDP stands for intelligent document processing. In addition to using proprietary machine learning models to read both printed and handwritten material, it also uses AI to extract, sort, and manage structured and unstructured documents. With a claim of 75 percent to 95 percent automation for data entry and sorting out-of-the-box, Hyperscience features an AI system that keeps improving classification coverage by continuously looking for improvement opportunities. It also features TLS encryption, RBAC, and several authentication options like LDAP, SAML, and OIDC.
Accuracy is important when you’re scanning and storing thousands of paper documents. Hyperscience features a pretty neat way to achieve the desired accuracy before the AI model is smart enough to handle it autonomously. Like how you would train a person on the job, the platform features a “human-in-the-loop” module that can be used until the desired level of accuracy is achieved. However, what’s really interesting is that a separate neural network is continuously trained based on what it processes so that it can reach the human accuracy level as soon as possible.
Next on our list is another startup using machine learning and artificial intelligence to automate document processing. Based out of Germany and Poland, Hypatos raised $11.8 million in its seed funding round and uses artificial neural networks to extract machine-readable information from documents, both structured and unstructured. This shows how complicated simple human traits like the ability to make judgment calls, analysis, and intuition actually are, and it takes a combination of several technologies to mimic those abilities.
Hypatos combines natural language processing, artificial intelligence, machine learning, artificial neural networks, and word embedding, and calls this approach CPA or cognitive process automation. While Hypatos claims its approach is superior to CRM, ECM, or RPA bots that can only mimic simple, repeatable business processes, it also offers support for users who don’t want to change their entire platform. Such users can still leverage this particular ML brand by using the Hypatos API to increase the level of automation and efficiency on their current platform.
Unlike the Hyperscience AI that puts a lot of energy into data extraction, Hypatos focuses on processing financial documents at speed and efficiency. Hence, the focus of its ML model here is not on deciphering human handwriting but rather on acquiring in-depth knowledge and understanding into the world of banking and finance and financial documents. These could be ledgers, checks, purchase orders, loan documentation, insurance claims, delivery notes, or a number of other possibilities. Hypatos claims a 90 percent cost savings with its automated document processing suite and options for customized automation.
Next on our list is a New York-based startup that, in addition to raising over $24 million in its Series B funding, has a unique take on document processing. Unlike other organizations trying their best to eliminate the human element, Ocrolus combines man and machine in the form of AI and crowdsourced “human intelligence.” Similar to how Hyperscience has a “human-in-the-loop” mode to train for accuracy, Ocrolus believes it’s always better to have a human involved and administers human quality control that, as we already mentioned, is crowdsourced.
Like Hypatos, Ocrolus is also focused on fintech, and in particular, on using AI to analyze financial documents like bank statements, invoices, pay stubs, tax documents, and more. While Ocrolus is used to extract, validate, and structure data from various sources, including cell phone images of any quality with 99 percent accuracy, Ocrolus+ can connect to other digital sources to process financial documents. Use cases for Ocrolus+ include loan underwriting, know your customer processes, mortgage financing, invoicing, and backfile data extraction.
What’s really interesting about the Ocrolus approach is that instead of spending huge amounts of money on trying to replicate human intuition, they pretty much embrace the fact that human judgment is supreme. While machines need vast amounts of data to come to a conclusion, people are a lot better at making decisions when limited information is available. We’ve always known that it’s human and machine intelligence together that make the winning combination, and that’s pretty evident from Ocrolus’s 99 percent accuracy rate.
sign of diligence, you can bet they need some automation in their lives. Founded in Singapore, and now with headquarters in Japan, Silicon Valley, Vietnam, and Taiwan, this Tokyo-based startup called Cinnamon is focused on “spicing-up” repetitive tasks. Miku Hirano, co-founder and CEO of Cinnamon, was quoted as saying, “Japanese people work too much,” and further explained how repetitive tasks can be “troublesome.”
Cinnamon features a Flax Scanner that extracts only the most important or relevant information from lengthy documents instead of ingesting everything it sees. It also uses a natural language processing (NLP) engine called Lapis to extract key points from all kinds of structured documents like driver’s licenses, passports, and bank statements. It can also extract data from unstructured documents like invoices and complex documents like contracts and tenders. It also features a chatbot called Scuro that, in addition to being NLP capable, also utilizes an intent classifier and responder module to communicate with and give recommendations to the user.
Big opportunities for document management startups
Startups are all about pushing the envelope, and while the current situation in the world has caused a lot of grief, it has created amazing opportunities for others. Hypatos, in a recent press release, stated that COVID-19 has caused an “uplift” in demand, much like VPNs are straining to keep up with the work-from-home scenario. While every startup on our list has a slightly different approach to document management, it’s definitely “uplifting” to see AI being used to defeat drudgery and repetition.
Featured image: Pickpik