It’s only been a few years since IBM's Watson technology beat Jeopardy! champions on US national TV.
Now Big Blue's supercomputer is taking on an even tougher challenge: coming up with trip-planning advice that travelers find relevant.
IBM is licensing its Watson natural language-based cognitive service to new startup WayBlazer, a B2B service that will power the consumer applications of third parties such as direct marketing organizations (DMOs), hotel chains, airlines, and travel publishers.
The Austin-based WayBlazer claims to digest unstructured textual information and then provide advice and transactional recommendations to travelers.
WayBlazer says it can, with the help of Watson, use natural-language information to answer difficult questions like, "I'd like to go to a four-star beach resort in January with my wife and two kids, and I'd like activities for two kids plus recommendations for good restaurants."
WayBlazer, currently in beta, has an acting CEO in Manoj Saxena, who is also managing director of The Entrepreneurs' Fund, an early stage venture investor whose IV fund is backing the project.
WayBlazer is partly the brainchild of serial entrepreneur Terry Jones, who was the founder of online travel agency Travelocity.com and the founding chairman of metasearch giant Kayak.
They have huge help on the sales front.
IBM's Watson sales representatives tasked to travel and transportation companies have already begun selling WayBlazer products to third-parties. The reason: IBM stands to gain from increased licensing fees of Watson technology.
WayBlazer created a Vine for Tnooz to illustrate its premise:
Answers to the hard questions
The B2B company's first customer is the Austin Convention and Visitors Bureau. Here's what it's doing, in this case -- though the results are not yet live:
Austin pops in code onto its desktop and mobile websites. The code intercepts a natural-language query made on the sites by a traveler.
WayBlazer also sucks in all of the information from Austin's database regarding about 1,400 activities. It builds concepts to distinguish the types of information that's relevant for each category of activity (restaurant, ranch, etc.).
It analyzes the relations between the data points; and, for this client, responds with a formatted webpage that has, to start, a couple hundred thousand "insight panels".
The traveler will only see a handful of these insight panels, or results -- namely the ones deemed the most relevant to his or her query.
Q&A with Terry Jones, chairman:
Tell us how you founded the company, why and what made you decide to jump in and create the business.
Last year, I was invited to IBM’s headquarters and was asked to help apply Watson technology to the travel space.
I became thoroughly intrigued in the implications of what cognitive computing could do for the travel industry.
I also had the great fortune of becoming good friend with Manoj Saxena [pronounced man-OHJ sax-SEE-na], who for three years was the GM of IBM Watson. He and I determined there was an incredible business opportunity.
Watson is very different from the kind of computers I ran at Sabre or previous startups. It's a computer that speaks and understands natural language. It can make sense of unstructured data, like words on an internet website.
It's also a learning computer. The confidence level at which it can make conclusions, predictions, and recommendations improves with time, the more it analyzes.
It's pretty clear that the travel booking problem has been solved. But the dreaming and planning stages still need help.
The number-one source for travel advice is still real people, like friends or a live agent. Social, blogs, reviews website, etc., offer clues as to what to do, but they haven't cracked the code.
WayBlazer can change that, make planning a trip online as easy as having a conversation with a friend.
We add Watson's natural-language processing to a cognitive travel graph we license from another company in Austin, Cognitive Scale. That graph database allows us to relate concepts to each other -- people, places, times, deals.
WayBlazer then creates a B2B product that is in a plug-in-play mode for third-parties.
The traveler will receive a set of suggestions, each with a confidence level of, say, 80% accuracy, because it is based on 100 reviews on 50 blogs and 10 tweets.
Our product can also provide "insider insights" as well. If a traveler is going to, say, Austin Formula 1, it might say, "Turn 15 is where you want to be for the most action".
The results could also, if the client wants, link out to an e-commerce opportunity, so if the suggestion is a restaurant or a tour that the customer could buy, we'd link to that.
By the end of the year, we expect to be offering e-commerce opportunities.
We plan to be able to suggest not just that a user stay at, say, the Omni Austin by repeating the hotel's own promotional text. Our results would supplement that promotional text with other relevant information, such as that the hotel had 124 Foursquare or Swarm check-ins from race fans last year.
Airlines have rich data on their best customers, too, and our results could be filtered not just by where the customers are but who they are -- how loyal, and so forth.
What we've done, that Watson hasn't, is sourced hundreds of thousands of open-source images on the web, and then we can supplement results with visuals, such as images of tours-and-activities.
Size of the team, names of founders, management roles and key personnel?
WayBlazer is currently a team of 10 today... and growing fast!
Terry Jones, Chairman and Co-Founder; Manoj Saxena, acting CEO and Co-Founder; John Faith, Chief Product Officer & Chief Technology Officer; Nij Chawla, VP Product & Client Service; Patrick Duncan, VP Content & Business Development.
We are actively searching for a CEO.
WayBlazer is currently funded by its founders and The Entrepreneur’s Fund.
Estimation of market size?
All of the travel market. The online travel industry alone is a $1T industry!
Revenue model and strategy for profitability?
For DMOs, WayBlazer’s revenue model will be based on percentage of increased advertising, percent of increased commerce and software-as-a-service (SaaS) fees. For other clients, it will be similar, but also may include NRE for custom code.
Our target customers are DMOs, travel service providers like hotels, and travel publications. Our products are Cognitive Search, Cognitive Commerce, and Cognitive Insights and Advice.
For DMOs, it could be a straight SaaS licensing model -- a recurring monthly fee, let's say. Or if the DMO has advertising, we could take a piece of that revenue based on some ratio because we boost pageviews and clickthroughs.
If they want to do commerce, we could add hotels, attractions, in a shared-success model of taking a cut of the revenue.
On the commerce front, we think we can help hotels, too. Go to a hotel today and sign into its Wi-Fi network: You'll typically see a promotional offer to book a room. Well, you might think to yourself, "I'm already here, dude. I don't need that."
How about something that says, "Monday Night Football, two-for-one beers, come down to our on-site sports bar." Or, "here's a great tour you could take of town," where the hotel gets a commission.
Airlines could also extend their relationship with a traveler on the ground.
What problem does the business solve?
The most successful online products (with a few exceptions) have mostly focused on booking. Few have focused on planning and dreaming. Though many have tried, it has been a difficult space to crack open.
We think cognitive computing technology will allow us to actually give travellers advice rather than endless lists of clues as to what they can do at any given destination.
We provide insights-as-a-service to travel websites looking for a greater level of engagement with their clients.
We're starting with the US as a bounded problem. We built our first client's solution in under two months, plus a month of design time -- and we expect a faster turnaround time as we get smarter and work with new clients.
We don’t think anyone is doing exactly what we are doing with WayBlazer today. Many online travel companies are doing parts of this, but not at WayBlazer’s full capabilities.
Other companies, such as Hopper and Nara, have to use natural-language analysis to produce conversational trip planning results, but they haven't caught fire. Why do you think you'll go viral where they haven't so far?
Yeah, and there have been others beyond the ones you mention. I think, first of all natural-language search is important but not sufficient.
The real power here is the ability of Watson to ingest all this data, make connections in a relevant way, and produce insights around them. Natural-language is just one of the ways for that to be accessed.
An example may show what I mean. We're working with a customer who wants a mobile solution for a traveler.
If you are a traveler whose flight has just landed in Atlanta, and you hit the "insights for today" button on the client's travel website, the site could give personalized, location-aware insights on what to do that are relevant without involving natural-language at all.
We could tell the user, "It's going to be raining this afternoon, and there's an art exhibition on until 5pm at such-and-such a place."
We also stand out from rivals in not not curating any of our results with human help.
How did the initial idea evolve and were there changes/any pivots along the way in the early stages?
We aren't quite old enough to pivot…yet. But we have added several client types that are interested in the product including magazines and visualization sites.
A couple of large travel publishers have reached out to us. We didn't reach out to them. And we're working with them to see how we could improve their content, search, and commerce opportunities.
Why should people or companies use the business?
Companies should use WayBlazer to increase visitor engagement through the use of our unique natural language capabilities and database to provide insightful answers to their visitors’ questions.
Here's a truly simple solution. One customer wants to use our tech to power the frequently-asked-questions section of its website. This is a use of our technology that's like killing a fruit fly with an atom bomb, but they want to start small.
They want to enhance the FAQ answers by pulling in every Q&A about the company on the web -- on Facebook, Twitter, etc. -- because our customers may have better answers than we have.
We can combine all of the answers into insights that the travel company couldn't on its own.
We want to get to the point where a client's systems could make a call that gives us a natural language query plus the customer relationship management CRM score of the customer, and we could send back some answers.
We're not there yet. But by next year we hope that, if a client just wants to take its data and our data, we'd do the data ingestment part, and we could create a result that would be effectively an API call.
Once we develop API-level solutions, we can help power a travel brand's mobile apps, too. It's on the product roadmap, but we don't have a launch customer.
While small DMOs might feel safe to partner with WayBlazer, some larger companies -- like a global hotel chain -- might be leery of partnering with you. Why should companies trust sharing their data with you?
There are privacy standards now about data. Watson's technology was built first for the medical sector, where there are intensely detailed rules about data security.
We can honor data sovereignty. So, where the data started and who's data is it and who gets to see it -- that's all built into the graph.
We can do a split-premise installation where your data never really leaves your data center. We just link it together. Your data is yours. Ours is ours.
There's also a political issue here. Big companies have internal R&D teams and data science teams that wouldn't want to get upstaged by you. But small companies may not have the resources to pay your license fees. How will you escape that bind?
The problem can be real, but not with every customer. I've done a lot of B2B selling in my career. There's always objections to outsourcing, and we just have to overcome them.
The coolest thing we have in that regard is that IBM is doing a lot of the selling for us.
A side note: For a large enterprise customers, we have access to Cognitive Scale's digital garage. The facility can bring in a big customer, they bring in people, we bring in people, we scale up the cloud quickly, they can have a demo running in 12 hours, and in 10 days they leave with a working prototype.
It sounds like a custom operation for each client. How is this scalable, packageable, repeatable?
No. Ingesting data is a repeatable task, it's not custom. We're more on the scale of, "what color paint do you want on your car?", rather than, "do you want a custom car?"
We have a canned set of output in a flexible user-interface. It's not as bespoke as PowerBuilder, but it's pretty flexible.
I don't want to be a custom job shop. But if we think one-off custom requests, which we are selective about, could take us into large markets, we'll take 'em.
It still seems like the client has to do a lot of work on its end to present the results in the branded way it wants. Right?
That depends. We have a pretty sophisticated way of presenting our results, and customers are liking it so far.
They're extended pages that look slightly different from their standard webpages. In the case of Austin, when there's a natural-language query, we're intercepting it and serving our own page back, which they say is fine.
Yet if a customer wants us to integrate the insights into what they already have, then, yes, it make take more development on their side.
If the traveler makes a request about a buddy's trip in the autumn to Austin, WayBlazer can suggest (in a white-label format), here are 27 things to do -- Austin City Limits, a UT football game, a fishing trip, etc.
Austin's developers don't have to have any special expertise. It's just a database call for Austin's own user interface.
Today, Austin's search function gives them a stream of links. Hopefully we'll get to the point where we're delivering a stream of insights, which can be presented in various ways that are up to the client.
Hell, when we built Travelocity, we were using obsidian knives, there were no development tools. Today there's lots of ways to solve problems. I'm not worried.
But for every additional effort on the client's end, they'll have to be a match in service on the WayBlazer end, to help guide them through the process and execute, right? At a basic level, even just to get the results to appear in a way that matches a brand's look and feel?
Sure, but again. Our goal -- and we're not there yet -- our goal is insight-as-a-service. To the extent that you give us a search query along with a persona about the customer if you know that, we serve back results in a manner you can incorporate without any specialized domain expertise.
What is the strategy for raising awareness and the customer/user acquisition?
Since our initial focus is on B2B, it will be direct sales and attendance at trade shows. As we are an IBM partner, they will also be helping create a demand for our product.
I've already joined the IBM sales force in pitch presentations to C-suite executives, before we formally launched.
Where do you see the company in three years time and what specific challenges do you anticipate having to overcome?
We need to continue to aggressively grow our database, constantly increase the level of personalization and deepen the sophistication of our graph database and IBM Watson.
I think in three years’ time we will have helped many sites increase the level of advice they give to their customers and will have found some incredibly special and unique ways to leverage this tech for travelers.
However, all that said, the market is going to tell us what it wants. If I've learned anything about startups over the years, it's that where you start isn't where you end up.
What is wrong with the travel, tourism and hospitality industry that it requires a startup like yours to help it out?
It is simply too difficult to get good travel advice today. That is one of the reasons that travellers search over 20 sites on their journey to plan their trip.
I’ve been in the travel industry from 45 years, and customers seem more frustrated than ever trying to plan a trip.
What other technology company (in or outside of travel) would you consider yourselves most closely aligned to in terms of culture and style... and why?
We are building our own unique culture – one that is vibrant, tech savvy, fast-moving and incredibly focused on the customer and the market.
Which company would be the best fit to buy your startup?
It’s a little early for us to be thinking about an exit! I think we are a good fit for travel technology generalists, and may be an even better fit with those looking to dramatically improve the search experience in travel.
Describe your startup in three words?
We needed one extra word: Insights as a service.
How long do you want to be doing this, Terry?
This is a blast. I don't have to do this. I want to do this. And I find it fun, given that I'm carrying a Medicare card. I'll be very involved for the first couple of years, as long as I'm adding value.
I'm doing it because I think this is revolutionary computing solving an interesting problem.
Until now, trip inspiration has been the part of the transaction funnel where travel startups go to die.
But unlike many startups, WayBlazer has a lot going for it.
It has the computing power of Watson, the open graph technology it licenses from Cognitive Scale, IBM's sales support in opening doors and co-selling, and the entrepreneurial moxie of Jones and Saxena. All of that could combine into a potent force, especially if venture capitalists sign some impressive term sheets with it.
IBM is spending about a billion dollars in enhancing Watson technology. It has also created a $100 million fund has publicly committed to give side-by-side investments in companies that harness Watson and meet certain criteria.
If anyone can crack the code of trip inspiration, it's the WayBlazer team.
(For more context, see Tnooz's earlier interview: "How to innovate: Lessons learned the hard way by Travelocity founder Terry Jones.)
NB:TLabs Showcase is part of the wider TLabs project from Tnooz.