Whatever happened to IBM's Watson? It's been more than three years since it defeated the "Jeopardy" champions, and not much has been heard from it. There was a brief flurry of interest in January when CEO Ginni Rometty made a $1 billion pledge toward the development of Watson, but specific innovations have been lacking.
That may be about to change. IBM recently rolled out a new tool called Watson Analytics, a natural language cognitive computing service (that's the current term for artificial intelligence) that allows a company to interact with and interpret its own data. So a company's executives will be able to ask, in natural language, "What advertising medium is providing the best return for our investment?" or "What is the most efficient deployment of my sales force?"
IBM has announced that it is moving most of Watson's staff to a new building smack in the heart of downtown New York's hipster/programming scene in an obvious effort to attract ultra-cool 20-somethings to develop apps for the cause.
I could care less about a new building, but I very much care about cognitive computing, and you should care, too, because what IBM and a group of competitors are doing will likely change the way a lot of business is being done.
Cognitive computers don't think by themselves, but they are capable of learning. They can adopt to new information, they can interact, and they can propose solutions. And the technology is getting better very fast.
Several deals are being announced. Terry Jones, founder of Travelocity and founding chairman of Kayak.com, is launching WayBlazer, a new travel company built on Watson that will allow travel agents and consumers to interact with Watson using a natural language interface.
Read More
Consumers and travel professionals will be able to talk naturally with WayBlazer about travel plans; it will make recommendations on flights, hotels, places, destinations and offers.
CaixaBank in Spain is teaching Watson Spanish so when clients call with banking and investment questions, branch office advisors can bring in Watson to answer complex questions immediately, rather than calling in experts.
IBM is also rolling out new apps using Watson in health care and sales management. For example, GenieMD's app will allow patients to ask questions in natural language, such as, "What are the signs of a stroke?"
It's all part of an effort to build on the breakthroughs with Watson, then combine it with IBM's long-time strength in analytics software, and then ... here's the critical part ... deliver it all as a cloud service. Watson is a critical part of IBM's plan to develop cloud service, an area in which IBM has lagged.
I spent an hour with Mike Rhodin, senior VP of the Watson group, on Monday night at the new HQ at Cooper Union Square. He seems like the perfect guy to sell artificial intelligence to the masses: He's intense but not too intense, nice suit but no tie, and there was no hint of computer jargon in the whole conversation.
Read MoreYou may soon getfinancial advice from a machine
One thing's for sure: IBM needs a revenue generator. Revenues have declined every year since 2011.
What I wanted to know first was, how much buy-in has there been from the top at IBM? Rhodin says he has complete support, and as evidence he points out that the entire operation is being run as a start-up within IBM. He reports monthly directly to Rometty and a few board members.
He refuses to be drawn into any discussion of how big a business Watson can become. He won't talk revenue projections, noting only that Watson is currently "part of IBM's $20 billion analytics business." He noted that Deloitte estimates the cognitive computing market will expand in five years to $50 billion in the U.S. alone.
There are several headwinds for IBM and Watson:
1) Too big to succeed, and potent competition. IBM is big. Really big, and there's some doubt it can pull off something like this. The current belief is that the only way any technology could advance is if it were invented by a 19-year old in his basement.
Read MoreArtificialintelligence could end mankind: Hawking
Maybe. But that's an easier call to make for social media than it is for artificial intelligence.
Invention does not always come from kids or small companies. IBM will have 2,000 people working on Watson by year end.
And those people create ideas. And patents. And they have moved to lower Manhattan to enlist all those kids to write apps for them.
And the competition? It's not so much a 19-year old, it's Google Now. Heard of it? Not exactly a start-up.
Read MoreIBM cuts payfor workers it says can't keep up: Report
2) Execution with speed. Right now, there's not much competition in this space. They are one of the few players that have a product. But in a couple years, there will be a lot. So there is a sweet spot right now, and if they don't execute right, they will be swamped soon with more nimble competitors.
3) Focus. It's hard enough to build a product in one business like medicine; it's a lot tougher to build a product across many different businesses, which is what Watson is trying to do. Executing sales in multiple businesses is never easy.
4) Buy-in from the developer community. IBM has just started to commercialize Watson. It needs to become a platform and get third party innovation and apps. That's why they have moved to downtown New York City.
For those of you not versed in what Watson is doing, here's a brief primer.
Professionals in all walks of life are being inundated with data. They can't keep up. Doctors, lawyers, academics, it doesn't matter.
Here's a mind-bender: Ninety percent of all digital knowledge has only been created in the last two years. Two years.
The response to more information is increasing specialization in all walks of life. The result is that more information is being locked up in silos that only a few can understand or have access to.
As a result, information has now become the currency of any trade. But how does anyone stay current anymore? How do you know if your knowledge is still relevant? How do you navigate through it all? How do you unlock all this information and make sense of it?
Watson is about Big Data. It is about ingesting vast amounts of information on specific subjects—medicine, law, travel, retail, metallurgy, oil and gas, whatever—and allowing a user to query the data to look for patterns, assist in a diagnosis, assist in a legal argument, make a decision on where to drill for oil, almost anything.
Let's take an example. Watson is initially being tested as an aid to doctors to more quickly and accurately make diagnoses. Why medicine?
1) Medical researchers can read at most a few hundred medical papers a year. Watson has ingested all 23 million medical papers in the National Library of Medicine (MEDLINE).
2) Medical errors are now the third leading cause of death in the U.S., according to IBM.
Watson is designed to interact with the patient's medical history, as well as information from the doctor's medical exam. So if a patient came in with a difficult diagnosis, the doctor could query Watson, which would compare the symptoms against a vast body of medical knowledge to produce a series of possible diagnoses. This is particularly valuable when dealing with rare diseases where the doctor is likely to have little knowledge of the disease or its symptoms.
The point is this: There are connections that exist to make a proper diagnosis, it's just that in many cases they are not obvious. But the connections are there; the right lines just have to be drawn.
Think of another area: crime. Cold cases. They are usually solved because someone has found an obscure connection with some person, object or event. But this is what expert systems excel at—finding an obscure connection.
And, finally, to answer the question, "What does Watson look like?" the answer is, like a pizza box. Well, it's about the size of three large pizza boxes.
Of course, no one is going to "deliver" Watson. It's not a single computer. Watson will be delivered as a cloud service.
PROGRAMMING NOTE: Watson head Mike Rhodin will join me and the rest of the CNBC "Squawk Alley" team at 11:45 AM ET.