As a company, the name ICG Solutions implies nothing.
It doesn't hint at an angle, industry and/or motive. It is amorphous, just as its technology is intended to be, according to founder David Waldrop.
ICG Solutions, based in Potomac, Md., boasts just one product: a powerful data analytics engine that can be fed millions of individual data streams—anything from social media posts and geospatial data to open-source internet search information and cybersecurity threat intel—to identify connections between specific "events."
This cloud-based software is called LUX.
Over time, LUX can find correlations among these events. Those discoveries can ultimately reveal trends, identify specific individuals based upon certain behavioral patterns and offer insight into public sentiment towards a person, policy or other topic. Naturally, the use-cases for such technology are diverse.
"LUX has more than 30 types of ingest plug-ins with the capacity to add new ones in a matter of hours. Estimating a total number of feeds across the install base is difficult (i.e., we have not yet encountered a limit). To date, LUX has ingested and processed data from more than 100 distinct sources," said Waldrop, including live news via GDELT, web intelligence from Bright Planet and maritime shipping activity through ExactEarth.
In 2015, 12-person ICG served 9 clients. In 2016, the company is targeting 20. Roughly 90 percent of current business comes from federal clients.
Today, the firm is out raising a Series A round to expand in size and scope.
LUX is sold in two distinct models, one for the defense space and another intended for commercial use. One of the inherent differences between each version lays in the types of data feeds plugged into the engine. Obviously, an organization like the Pentagon has access to different realtime data than a brand like Pepsi or McDonald's.
Waldrop's ICG business, at least to date, has revolved around "three-letter agencies" in the defense space, but preliminary meetings with the RNC and now-defunct Jeb Bush campaign show that others are in the market for a big data analytics engine with visualization capabilities. D.C.-based thinktank the Center for Strategic and International Studies is a potential customer while negotiations are underway with cybersecurity firm Fortinet, according to Waldrop.
In an interview with DC Inno, Waldrop shrugged off the comparison between his firm and market leader Palantir Technologies—"we're just different."
In recent months, the prevalence of big data technology in the world of political campaigns has proven significant. Whether it's been a revelation that Ted Cruz's campaign employed Cambridge Analytica, an audience targeting firm typically used by private companies, or even the introduction of such a concept in Netflix's hit show House of Cards, the idea that software is playing a role in politics is simply becoming an assumed reality.
ICG Solutions specializes in streaming real-time analytics to help organizations rapidly mine and correlate massive volumes of data.
In terms of an election, this would mean, among other things, running social analysis; ingesting mass volumes of publicly accessible posts, statuses and other comments on social media platforms like Facebook and Twitter to get an idea for how people are reacting to a candidate. In such a case, all of this data is open-source/public; the software cannot pull private information on its own.
In a demo of LUX, Waldrop showed how the software was able to find specific words, phrases and expressions that were becoming increasingly popular on Twitter as it related to politics. Here's a look at the data through a keyword visualization—the larger the word, the greater frequency it appeared on social media during the election cycle in December 2015:
"LUX’s real-time architecture enables analysis of multiple feeds of dynamic data in without first having to store or structure the content. Analysts themselves can modify and deploy analytics directly to the system in minutes through a user-friendly interface. These two innovations allow business users to transform information into insight–when right now matters," explained Waldrop
In another example (pictured below), Waldrop showed DC Inno how LUX's data analysis interpreted a live primary debate between former secretary of state Hillary Clinton and Sen. Bernie Sanders.
In this image, the visualization illustrates the popularity of specific policy issues in text on social media. The below line graph, with its peaks and valleys, provides both a reflection of the debate and a window into what topics were resonating with people on social media (over a 1 hour period during the debate). A spike in alert counts could, for example, may be the result of a heated exchange between the two career politicians.
LUX's results and more specifically, its ability to provide insight is dependent on a series of rules—some of which that are already written and others not. In other words, there needs to be a structure, algorithm or definition to how the system is able to, for example, decipher when a social media account is owned by a female or male, their age and their political affiliation. To do so, it's a complex, multi-level process which layers speech/language recognition tech and attributes certain words to either negative or positive connotations.
In essence, this all translates into an engine that is highly customizable but that also relies greatly on the expertise of a user to understand exactly what they are looking for. To borrow from a metaphor: the task of finding a needle in haystack becomes infinitely more difficult when you don't know what a needle looks like.
A simple question like "Is Donald Trump liked by millennials" wouldn't fly on LUX. First, one must set rules in place for the sort of language you perceive as positive and equally, how you quantify a millennial. Only after proper and careful preparation, will the system be able to give you a precise answer—which brings to light another question: what are you really hoping to find out?
I asked Waldrop how LUX—support by some template, pre-fixed rules—would be able to gather from my own Twitter profile based on my activity over the last several hours. The picture (below) shows that the system realized I was male based upon my name and suggested my likely party affiliation to be democratic based on a few of the people I follow. The software also looked at one of my recent retweets, which it believes suggests I am anti-Donald Trump—spoiler alert: true.
Founded in 2011, it may be fair to suggest that LUX's use base—even while it's UI is easy to navigate—may be limited. But for those who are able to control it, the information it provides is objectively impressive.
Where ICG Solutions shines, Waldrop described during an in-person interview, is in its pure ability to ingest massive volumes of multi-source, realtime data—all at once, without crashing.
