Spending a day with IBM’s Watson

As an IBM alumnus (but no longer a stockholder) I’ve gotten pretty used to seeing the company do things a certain way. And then I attended a day-long “Watson at Scale (aka Ecosystem 2.0)” event on October 7 and had a lot of my old notions upended. Watson, of course, came to prominence when it won Jeopardy in 2011. Immediately after that IBM began experimenting with a select number of industries (Healthcare, Travel and Retail) to demonstrate proofs of concept and learn what works and what doesn’t. Beginning in January of 2014, Watson expanded dramatically and is now covering 26 industries. IBM proclaims that Watson is the harbinger of a new era of computing, what they call “Cognitive Computing.” There’s just too much information being created today for any single person to digest; Watson aims to “amplify” experts’ capabilities. Doctors, salespeople, and wealth managers are but a few examples. IBM says there are four key attributes to understand:

• Watson understands natural language (computational linguistics).
• Watson is a voracious reader
• Watson provides recommendations with confidence levels
• You don't program Watson, you teach it

Mike Rhodin, the IBM SVP who leads Watson (under an unusual board-governed structure), described the key insight about Watson: it’s not that Watson gives answers, but rather that it generates hypotheses, gives confidence intervals around those hypotheses, and provides evidence trails. It does this by ingesting enormous amounts of data, being taught by humans with a series of questions and answers, and then learning on its own as it proceeds. The more data it has, the better it performs. Watson does a better job providing recommendations for people when it knows something about them. A salesperson will sell differently to an introvert than an extrovert, as a simplified example. Watson can generate a personality profile on the basis of a person’s twitter feed or blog posts – I’m not sure how accurate it is, but the concept alone is pretty startling. Terry Jones, an entrepreneur previously associated with Travelocity and Kayak, introduced a new company called WayBlazer that uses Watson’s technology. It aims to be able to answer queries like, “I want to go on a golf trip with my buddies in October,” or, “Give me an itinerary for Costa Rica in May with my two kids.” Watson might ask clarifying questions, and then would come back with recommendations. The prototype is currently in place for Austin, but a service like this highlighted clearly what Watson has the potential to do, if implemented successfully. Another lightbulb for me was Terry’s description of Watson as a liberal arts major, not a math geek. It could do airline pricing optimization, but that’s not what you’d buy it for. WayBlazer is but one example of the ecosystem that Watson is building. Realizing there’s a shortage of skills in Cognitive Computing, IBM has teamed with ten universities to offer courses on the subject; this fall all of the classes were oversubscribed. From a standing start in January, IBM has about 100 partners and expects that to continue to grow. Watson had 1 API in January; it now has 8, with more than a dozen in development. IBM may have finally figured out how to execute at startup speed under the umbrella of Big Blue. Watson’s interest in financial services is currently very focused. Insurance is one key space, particularly around underwriting. Wealth Management is the other key area, with risk and compliance being a third. You may disagree with Watson’s prioritization, but their intentional focus is spot-on – they’ve got to demonstrate some tangible successes before they begin to branch out. Based on different discussions, Watson’s revenue will come from four sources:

1. Consulting to investigate and establish what Watson will do for the firm
2. Priced products (e.g., oncology)
3. SAAS revenues from running Watson for individual projects
4. A cut of the revenue that partners earn from Watson projects

What’s ultimately different this time? In this new IBM (a place the company has been forced to by intense competition), Watson:

• Is playing the role of an ecosystem platform
• Is using partners to reach consumers, realizing that IBM’s strength is as a B2B company
• Has built a new physical space, reversing a trend of selling real estate and having employees work remotely
• Is not trying to do this on the cheap
• Is focused on just a few areas

What does this mean for banks and financial services firms? The IBM take is, of course, that you’ve got to be exploring Watson or you’ll be hopelessly behind. That’s overly broad, but I recommend that firms at least get up to speed on the potential of the technology and see whether it can apply to them. We’ll have to watch to see if Watson carries through on its promise, but efforts like this are a necessary (if not sufficient) first step in the right direction for IBM.