The Evolving Nature of Data Management

The meaning of words and phrases change over time. Take for instance the word ‘awful’, which used to mean ‘worthy of awe’, but has evolved into its inverse, and now largely means ‘bad’. ‘Nice’ used to mean ‘silly’. Fizzle used to mean “silent flatulence.”  This evolution even occurs in technology. The term “data management” has evolved faster than the guppies of Trinidad, which are supposedly really fast evolvers.

The term ‘data management’ crawled out of the primordial technological soup back in the 1980’s as technology moved from sequential processing to random access storage, making the management of data imperative as it was stored in multiple locations.

As time passed, and the quantity and value of the data grew, the term data management evolved into information management, which referred more to the business analysis of the data than the actual physical location of it. The evolution of the term not only halted, but regressed with the popularization of the terms ‘big data’ and ‘data science’, which are defined as fields that looks to extract meaning and value from the underlying data.

So the term data management has come to mean something quite different from its nascent definition. It is no longer about storage, compute and hardware, but seems to be firmly associated with the analytics and the extraction of value and insight from the data itself.  

What I find confusing then is how certain modern “storage” companies have not recognized the evolution of the term, referring to themselves as data management companies in one form or another.  What most of them mean by ‘data management’ is that they backup data, archive the older versions of that backup data, and on occasion restore data from the backup or archive tiers. This obfuscation extends to every industry that touches data in one form or another. This impacts me personally because I work for a data management company called Promethium, and I am consistently explaining to people that we are not a storage company.  We are a company whose mission is to help enterprises find and extract the value of their data in seconds and minutes, not months, so less time is spent locating and making sense of the proverbial data swamp, and more time can be spent extracting value. In essence, we are a data management company.

So let’s evolve with the times. Let’s all agree on this distinction between storage and data management to help clarify and streamline the conversation.  If we can do that, wouldn’t that be truly awful?

How Can AI and ML Help My Business Grow?

AI is not coming. AI is already here, powering the tools and infrastructure that can turbocharge your business, but navigating the complexities of AI and ML can be daunting. So let’s break down a few of the barriers and challenges standing between you and the day your organization begins to benefit from AI based technology.

Let’s address the most often asked question first: What is machine learning (ML), and how does it differ from AI?

AI can be divided into two groups:

  1. Weak AI
  2. Strong AI

Weak AI is a technology that relies on algorithms to do a predetermined task. If you saw the new release of Jumanji, you might remember the NPC (non-player character) in the jeep that rescues the game players from a charging rhino. His responses to the players questions are limited to a predetermined set of answers that he regurgitates based on listening for certain voice cues.  

Strong AI differs from weak AI in that it enables machines to mimic human behavior. Due to the highly complex nature of its algorithms, strong AI can learn and adapt to a changing environment on its own. Most of the available AI today falls into the weak AI category.

ML is a subset of AI in which it uses statistical methods to enable machines to improve with experience. A machine is given access to a data set and is able to learn from it. In this case, a computing device given access to data about product shipping schedules and weather might conclude to build in 2 extra days on all shipments to Colorado and Utah due to potential snow storms in the area. Eventually, as the machine algorithm continues to learn, it may refine it’s shipping delay time to 1 day as it learns that trucks are more efficient than the machine’s original assumption at managing bad weather.

Why do I need ML based systems?

Petabytes of data are practically impossible for humans to manage manually. Parsing through mountains of data to understand what your data can answer consumes precious time, resources and the ability to focus on the true metrics that drive your business.  Having a bunch of data by itself isn’t going to help you succeed. Insight from the data is what is needed, and AI and ML have the power to unlock that insight.

Promethium’s Data Navigation System (DNS) has harnessed the power of ML to automate the entire analytics process. The DNS contains an intuitive interface that allows the user to simply ask the relevant questions. Data is then located and queries retrieved without the need for complex technical SQL query programming. Now the user can find the right data across different vendors, types and locations quickly and easily.

Building AI vs. Deploying AI.

Today, AI and ML systems can be quickly and easily deployed, meaning you don’t need to know how to build the systems, only how to operate them. For instance, when it is time to get in your car and drive down to the grocery store- (Wait, I hate going to the grocery store-let’s drive down to the grown up toy store instead)-you don’t have to know how to build a car. You just have to know how to operate a few simple controls on the car sitting in your driveway. Promethium has been built to integrate into your environment quickly and easily, and the simplicity of our GUI means you will be up and running in hours, not days or weeks.

AI and ML Are Only As Good As The Data They Are Trained On.

The data sets that train these advanced systems are absolutely huge, and they have to be in order for the machine to learn from the many different scenarios it will encounter. Unless you work for Google, Amazon, Facebook, etc, it’s probably unlikely that a single company or business unit can generate enough data for ML to be effectively trained.  Promethium’s SaaS solution leverages the Data Network Effect to train on several petabytes of data across its pool of customers. Those results are then verified to make sure that the algorithm is providing the most accurate and informed insights about your data across a wide array of data models.

Only large enterprises need AI OR ML.

This is not necessarily true. Even small and medium size businesses can benefit from extra insight into their data. Knowing what kind of questions your data can answer can take weeks or months parsing through data sets and tables. AI or ML can dramatically reduce the amount of time you spend searching for the right question, and move quickly to generating insights. Promethium’s ML based software gives any size organization the ability to contextualize their data, so they can so they can arrive at actionable insight quicker.  

Want to learn more about how Promethium can help you uncover the meaning of your data faster? Visit us at

How MLaaS Will Benefit Your ML Integration

In machine learning, one size does not fit all. There are many different types of algorithms to solve many different problems, especially when it comes to predictive modeling. And lots of ML based companies, and the venture capitalists that back those companies, tout their algorithm as ‘best-in-class’, claiming that their algorithm cannot be separated is what’s needed to solve all problems. At Promethium, we see things a bit differently.

Getting a Ph.D. in ML is not the only way to dip your toe into the ML pool. Machine Learning As A Service (MLaaS) allows you jumpstart your ML initiative without a Ph.D. MLaaS represents the commoditization of ML algorithms for some of the more standard applications such as image recognition, predictive maintenance and financial analysis. MLASS gives developers the ability to build, train and deploy ML powered applications quickly so you can get to production quicker with much less effort and lower cost. Amazon Sagemaker, Google Tensorflow, IBM Watson and Microsoft Azure are some of the leading end-to-end ML platforms in this space. MLaaS is still in its infancy, and whether it develops into a successful business model has yet to be determined, but the writing on the wall is clear. As top ML applications go mainstream, deployment options will move towards standardization to simplify and streamline the costs of adoption.

Because of this trend towards the commoditization of ML algorithms, Promethium built its  data context solution with the ability to be decoupled from the underlying algorithm. It doesn’t mean you have to purchase our software without our algorithm. You can and we believe in many cases enterprises will do this as a plug and play solution. But you don’t have to. IT may already have integrated a leading MLaaS offering throughout its organization and might not want to add additional, potentially conflicting algorithms to that universe. In that case, Promethium can sit on top of the ML platform, working seamlessly with your algorithm of choice.

About Promethium:

Promethium’s technology has been created from the ground up to increase productivity and revenue by simplifying and streamlining the process of Business Intelligence.  Just as Google Maps have simplified our personal lives, we aim to simplify the lives of our customers when it comes to analytics. Just ask your question and leave the rest to Promethium.  When everyone in the organization is looking at the same map, collaboration becomes infinitely easier.

To find out more about Promethium, visit us at, or follow us on twitter at @promethiumi