Artificial Intelligence:  Solution to Everything or Just Snake Oil?


This blog first appeared on


You’ve heard it many times — “We are an AI company.”

It brings back memories of other hypes. Cloud, Software-as-a-Service (SaaS), and even Object-Oriented Design (yes, I’m that old!). B.Yond calls itself an AI company. Are we just the pot calling the kettle back?

The reality is that AI has been around for decades. Many core algorithms have been well-defined for equally as long. The first Neural Network algorithms were conceived back in the 1940’s. The Random Forest algorithm was first described in 1995.

So, why all the hype?

Well, for the first time, we have the distributed and connected computing power (think “cloud”), the memory performance (think Solid State Disks — SSD), and the network speeds to do AI in real-time, and at scale. Ten years from now, we’ll no doubt look back and laugh at how primitive everything was.

But there is no doubt that we are at a stage where technologies and applications are precisely aligning for the industry to make a leap.

Some call this an “inflection point.” Renowned and respected AI authorities like Andrew Ng claim that AI is the new electricity. His point is that, as soon as electricity was made widely available, exceptional innovations and great companies were born. Electricity was the commodity that fed new ideas and improved life. If you accept that analogy, you must accept that AI is a commodity as well. The innovation lies in how you apply it.

Is AI the solution to everything? Of course not! You can’t take any problem, sprinkle some AI fairy dust over it, and hope it will work. Neither is it simply snake oil hoping for the placebo effect to kick in.

Properly applied, AI will solve old problems and, in some cases, enable a new world of connectivity. At B.Yond, we believe that two aspects of AI technology development make all the difference:

  1. The Application of AI.

  2. Networked AI Functions.

The Application of AI addresses the current state of the art (or science). While computing, storage, and networking technology advances are undeniably impressive, they still only handle point problems. Most AI models, therefore, deal with a well-defined issue.

Nothing wrong with this at all. We are developing a framework that makes system resources available to AI models (and any other logically encapsulated software). This includes data ingestion, data lake, generic data models, dynamic network topology tools, and other resources that simplify and accelerate the development of business value for our customers.

Networked AI Functions are where point problems become building blocks to address larger business opportunities. By allowing discrete AI functions to announce themselves, subscribe to output from other functions, and publish their own results, you can build a new type of neural network at a higher level. Such an AI network is stochastic (that is, random) by nature, and is better equipped to deal with the ad hoc, infinite combinations of events that go on in a virtualized network environment, such as emerging mobile telco networks.

In fact, we believe that current Operational Support System (OSS) models that rely on rules-based and reactive paradigms are incompatible with emerging mobile networks. Instead, a predictive and proactive paradigm, coupled with closed-loop automations is required. This will be the only way to manage the dynamics of a networked cloud at scale. This is where applied AI on a networked and open platform will make all the difference. These are indeed exciting times to be mining the AI commodity!