Artificial Intelligence in 2016

Artificial intelligence (AI) is a very hot topic these days. Just ask Shivon Zilis.

Who’s Shivon Zilis?

She is an investor at Bloomberg Beta, an early-stage fund backed by Bloomberg L.P. Shivon does research on areas of potential investment for her firm, and decided to analyze and define AI for her firm and her blog readers. She undertook this effort to help clarify this space because the technology is advancing too quickly to stay on top of it, but more importantly, as part of a software solution, it’s solving high value business problems for organizations in very clever ways.

Shivon, who prefers the term “machine intelligence” over artificial intelligence or machine learning, published her first machine intelligence landscape in December of 2014 to reflect the notable vendors expected to be leading the space in 2015. Her first landscape, called “The Current State of Machine Intelligence” covers many vendors that use machine learning and natural language processing to create solutions that use “intelligence,” and can be taught to perform many tasks that only humans could do a short time ago.

If you look in the Legal industry section in her landscape from 2014 (shown below) you will find a few of our competitors.



Shivon has just published “The Current State of Machine Intelligence 2.0” describing the changes in this space and the evolution of vendors. She describes how she spent the last year “chatting with hundreds of academics, entrepreneurs, and investors about machine intelligence” and focusing on the strongest innovation and vision in her new vendor landscape.

If you look at the Legal industry section in the new report, you will see Seal Software.



There are two main reasons that Seal is now shown in the landscape of leading machine intelligence (to use her term) companies. First, we are a leading vendor in applying and combining technologies such as Machine Learning, Natural Language Processing and LSI (Latent Semantic Indexing) to solve a very real business problem – knowing what data is held inside large sets of unstructured legal documents.

We are not an AI platform or foundational piece for someone else to use to solve a business problem – our technologies were built to discover and analyze contracts. In fact, we’ve run over 100,000 contracts through the Seal platform to train our system on what a contract is, and how to uncover terms, provisions, conditions, and incentives, some of which are not so obvious. In fact, we don’t just recognize the words themselves, but Seal’s power is in deciphering groupings of words to understand their meaning and intent. This capability allows us to highlight the differences between standard contractual language and non-standard language for individual contracts across the portfolio, effectively removing risk from an organization.

The second reason is the way we’ve used algorithms within the platform to reset the bar on accuracy and speed. Our system uses a combination of approximately 15 different algorithms working together to maximize the power of contract discovery, data extraction and deep analytics. This set includes what are considered “state of the art” algorithmic methods, alongside well-proven ones such as MaxEnt, LSI, JAPE and SVM, just to name a few. In fact, we are announcing a major addition to our algorithms in our next release that will put us even further ahead of the market.

It is certainly an honor to be listed in Shivon‘s vendor landscape of the leading machine intelligence companies, but it was the long, long hours and mountains of empty coffee cups in our Swedish development organization that got us there.

The real honor, and fun of working at Seal is applying these technologies to be able to tell a large multi-national organization that after a takeover 4 years ago, they were auto-renewing a lease of $400K per year (true story). Or to help another customer that was losing between 4% and 8% of revenues due to not enforcing Retail Price Index pricing increases that had been negotiated in their contracts, only because they didn’t know how to find them, and we could tell them exactly where they were and the specific terms of pricing adjustments.

That’s applying machine intelligence to real business value, which is what Seal is all about.