2019 – A Year of Digital Intelligence

“We believe that if men have the talent to invent new machines that put men out of work, they have the talent to put those men back to work. “

Before we start, let’s define what Digital Intelligence is to us :-

“The application of mathematical methods to data to provide humans and computers with insights, decision assistance and decision making”

“By 2020, $1,000 (£581) worth of computer will equal the processing power of the human brain,” he says. “By the late 2020s, we’ll have reverse-engineered human brains.”

— Ray Kurzweil (in 2005)

For some years now mathematical methods such as Calculus, Linear Algebra, Probability, Statistics, and Optimization via Graph Theory, Quantum Graphs, Entropy Vortex, and so on have been used on data to achieve Natural Language Processing (NLP), Machine Learning (ML) and Artificial Intelligence (AI) allowing computers to interpret and cognify information. This provides methods for signals to be extracted from noise and in the case of Deep Learning, a sophisticated form of Machine Learning, Cognition and insights to be achieved. In the case of AI, decisions made or recommended…

I have just finished reviewing this years applicants to the Fintech Innovation Lab and reviewed roughly 80 applications. I also checked with my colleague who reviewed the other 80, and it was consistent that nearly 75% of the applicants were using at least one of the techniques described above. In nearly all cases though these were not “Data Science Platforms” but rather these techniques are being applied in a vertical or sub-vertical problem domain to solve a specific business problem or problems.

At Sand Hill East we have started to observe that companies who were early adopters (circa 2010+) of these techniques in vertical use cases who have refined their platforms with significant client input during deployments have really started to gain significant advantage with the enterprise in the 2nd half of 2018 and” reverse-engineered” the jobs of some rote and some less rote roles to Kurzweil’s point above.

Specifically,

Lastline – Applying Machine Learning and AI to Advanced Persistent Threat or APT

Moogsoft – Applying NLP, Machine Learning and AI to Operations Management.

Untapt – Applying NLP, Machine Learning and AI to Talent and Skils

MSG.ai – Applying NLP, Machine Learning and AI to Customer Service

have all benefited from this uptick in interest from the Enterprise.

Gartner Hype Cycle for late 2018

The Gartner model above positions Augmented Reality close to enlightened use but leaves Deep Neural Nets/Deep Learning at the apex of the Peak of Inflated expectations… Whilst all the companies above have had some level of incredulity to believe that the purported capabilities of their products can really work in a complex enterprise, each of them have won clients by proving their platform on prem (or in the cloud) on customer pain points and problems.

They (and others we work with) are now at critical mass, having obtained reference-able customers and most importantly, have word of mouth to drive increased market awareness. They have all worked hard to reduce time to value from PoC’s or PoV’s which is the trick in achieving trust. The mathematical techniques themselves have all been adapted, sometimes onsite, to the specific use-case and in some cases novel techniques created from scratch. For example Vertex Entropy as a Critical Node Measure in Network Monitoring by Phil Tee et al.

Nearly all of the early stage applicants to this year’s Fintech Lab are walking in the footsteps of these early adopters but with market acceptance at all time high and robust math platforms (like Data Robot for instance) to provide a massive time to market advantage vs inventing the art. It is my prediction that the acceleration of innovation applying these techniques in verticals will also benefit from an acceleration in the robustness and completeness of the math, a lot of whose research is done in academia. This is a real example (much like NAG in the 80s when I was writing quantitative code) where academic breakthroughs in algorithms can literally be tested / back tested on business problems within days or weeks.

It is as hard to predict the future as it is to convince an Enterprise buyer of AI capability, but I believe when we look back 5 years from now at 2019, it will be seen as the year when the early adopters of NLP, ML and AI in vertical use-cases became uniquely differentiated from their traditional competitors, and Digital Intelligence started to become mainstream.

Andy Brown is CEO and founder of Sand Hill East and CTO in Residence at Fintech Innovation Lab. He advises Untapt, Lastline and MSG.ai as well as sitting on the board of Moogsoft, all of which are referenced in this article.