"India will adopt Watson faster as there are no legacy investments"

IBM has bet big on its Artificial Intelligence platform Watson. In 2011, Watson beat humans at the quiz show Jeopardy heralding a new dawn for cognitive computing (which is simulation of human thought process in a computerized model using data mining, pattern recognition and natural learning process). Though work on Watson has happened since 2008, IBM made it commercially available from 2013.
 
IBM which is facing challenges in its traditional businesses of hardware, system integration and services - with 20 consecutive quarters of revenue decline - has bet big on Watson and new generation sevices. IBM in the past has said it would get about $10 billion revenue from Watson related offerings by 2024.
 
Sanjay Brahmawar, is the GM and Head of Global Sales for Watson's Internet of Things (IoT) platform where they work with partners like Whirlpool, Honda, BMW, Visa, Kone and Siemens to transform their business models. In India recently Brahmawar spoke on a range of issues related to Watson and IBM's challenges and opportunities to Venkatesha Babu. Edited excerpts from the conversation:
 
What has been the progress since Watson was made commercially available in 2013? There have been concerns about the pace of progress which was expected in terms of revenue contribution. Is Watson technology for technology's sake or is there actual real on ground impact?
 
We started working on Watson about 10-12 years ago. We now have Watson touching the lives of a billion people. For us, this was an investment in journey that will become more relevant and widely used (both by) individuals and industry.
When we won Jeopardy, we had one real capability that was being worked which is natural language interface. We had Watson ingesting tons of data and we had natural language interface so that you could get answer to a question, start calculating the risks and decide on ways of making decisions. This was something that normal algorithms were not able to handle (at that time).
 
Then we were like where to apply this, what will be the application. For this, the first case was healthcare and you can see the amount of work we have done in health industry.  
 
We have done a few acquisitions and have also built Watson Health. We have also partnered with a lot of significant hospitals to not only to support the experienced doctors but also leverage Watson to assist doctors in remote locations. So, we see a lot of use cases in the healthcare adding up. If you look at IBM's progress over the last five years, our revenue related to cloud and cognitive solutions has grown significantly. When I came to India last year, there were no client examples. Today, we have around 8-10 live clients where we are doing stuff on Watson IOT.
 
Watson was never limited to healthcare, auto, or manufacturing. Have you seen similar progress in the other areas as you have seen in healthcare?
 
We have the capabilities and now it is just a matter of where to apply it. The first one is healthcare and second is the Internet of Things. IOT has been around for a very long time. Oil/gas industries companies have been collecting data about their oil and gas rigs. What is different today is they now have a lot more robust platforms to collect data from various systems. Another point is you can combine structured and unstructured data. They were collecting a lot of images, text, written stuff, videos etc. but had no capabilities of combining structured and unstructured data.
 
With cognitive we have got the capability to combine all these forms of data. The last is the ability to be able to provide analytics in a very simple form to the users, i,e, engineers, field service technicians, and customers etc. When you combine all of that, then you combine the cost of sensors going down dramatically, cost of network connectivity going down and cost of storage going down, (Watson's utility will be self-evident).
 
Is there a fundamental difference between what something like Watson offers vis-a-vis conventional AI and data analytics which all companies have been peddling for some time?
 
We say we are building cognitive capabilities, you can call it artificial intelligence, there is nothing different from that viewpoint. The only difference is the perspective with which you are building these capabilities. For us, this is a partnership between human beings and machines. We are building it with a thought about augmenting human beings, rather than replacing/supplementing them.
 
These principles are very important. We have three principles 1) customer's data is their data. We will never own our customer's data, just like some of other competitors. 2) Any insights we draw from our customers' data will be their insights. They might give us permission to (say) use, to replicate in the industry, but we will never own these insights. 3) Our principles about data safety. As we build our cognitive capabilities, it is important to have very clear set of principles.
 
Another thing that sets us apart is our contextual and industry understanding. We are not a generic search engine, we will never be a generic search engine. We will focus on domains like health, Internet of Things, within IOT on retail, manufacturing etc. Within say the Oncology domain we are teaching Watson the hierarchy, the terminology, taxonomy, vocabulary of domains. So, when customers talk to IBM about Cognitive, they can talk to us about Cognitive in retail, Cognitive in travel and transport.
 
Finally, we do not believe that IOT is a single person's play. It is a massive ecosystem. You have to find a way to create a trusted ecosystem. Therefore, in our Munich center, we have customers co-located with us- BMW, Siemens, Capgemini, Tech Mahindra; all co-located with us, putting their people to build solutions together. We do not have the expertise in all the domains. Our clients bring this on the table and then we add our cognitive expertise to it. We build together strong solutions for a particular segment.
 
If Watson is a standalone, independent offering, would it play nice on Amazon Web Services or an Azure? Historically, IBM had this policy where they tried to own the entire ecosystem. Is it plug and play kind of offering?
 
Our customers have varied infrastructure and cloud solutions. Not everyone is sitting on IBM cloud. First of all, we build our platform on open cloud Bluemix which is our solution, but we build it as open. So that, anybody can use it, build APIs onto Watson. We are building an ecosystem which is more integrated and collaborative with our solutions (primarily Watson runs on Bluemix).
 
The second question is how we offer our solutions for clients who are using a different platform than ours. I will not comment about the entire Watson. But for Watson IoT, we are looking at opportunities to offer it to clients who are using AWS right now. I will not comment on entire Watson, because we need to see how this progresses and how this develops, but for Watson IoT we are looking for opportunities, to be able to offer Watson IoT to clients on AWS.
 
The value to the client actually comes from the industry solutions and not the base platforms. The base platform is just collecting the data. The value comes from industry template, industry analytics, the cognitive on top of it. We are not there right now, to be able to say we will offer it to everybody but we are working towards it. We believe, to grow we need to have those kind of solutions.
 
What time horizon are you looking to offer Watson on any platform?
 
Within this year.  
 
So if I am a startup on AWS and I come to you to use Watson, right now it might be difficult?
 
As I said, I am commenting around Watson IoT. Watson IoT  is primarily a B2B, B2C play. It is not so much of B2C direct play.  Majority of our conversations are with clients across seven sectors - automotive, manufacturing, retail, electronics, insurance, travel and transport and agriculture. I see a lot of traction in the last two years. In those cases we have clients that have with their applications sitting on AWS or Azure, but they are keen to partner with us on IoT and cognitive solutions.
 
Are there instances (India or globally), where specific savings that can be attributed to Watson IoT?
 
It is not only about savings but also about adding value to the business model. The first example is Kone. Kone wanted to connect every escalator and elevator to their IoT platform so that they can do better maintenance, predictive maintenance, condition based monitoring etc.
 
They have 1.1 million elevators across the world. They are trying to connect all of those to our Watson IoT platform and solutions. They move about 1 billion people every day around elevator and escalator. So, they want to have people flow management. They wanted to use this data to be able to help facilities management. People who have to clean, touch, operate, they know where the congestion is and where people are spending, how much time and when, first and foremost they can know usage patterns.
 
Lastly, they wanted to shift their business model. They want to pay for flow of people and not for the assets. They sell escalators and elevators to different parts of the world and they do not own MRO everywhere. By offering this as a service, they do not pay for the assets, it is like we will charge them as the asset gets used. So they can include the cost of the MRO in the pricing.
 
Is there any dollar number which they have put saying that because of implementing the solution we have been able to sell more elevators/escalators or this has been our cost saving because we have been able to prevent break downs, ensure better people management?
 
For Kone, it will be the first financial year after the implementation. They have been in the process of documenting how many dollars they can save. They have been with us across the board right from the day they signed us. They want to be ahead. They are very bold with how they make their statements.  They want to clearly communicate that they are different. They expect 20-30% improvement in operational efficiency.
 
They want to grow their revenue in double digits when nobody in the elevator industry is meeting these ambitious goals as of now. Kone is a $7-8 billion Finnish based company. They do not want to become a $30 billion company but are looking at double digit growth.
 
Any other example which is number attributable?
 
We have Siemens. They have adopted our technology. They set the claim that they would save about $140 million in energy cost to their customers because they do building management systems. It is the system to be able to manage the efficiency of the energy use, whether it is around light, it is around heating, gas etc.
 
Honda is another global example. We now have many global clients. From my perspective they have not just invested in only for cost reduction. Although if you think of IoT, for us the benefits are in three buckets. 1. Operational efficiency, performance, cost reduction. 2) customer engagement, quality of engagement you do with your customer 3) new business models, Mainly moving from product to service. Most clients we are collaborating with are attacking all three buckets or one or two of these buckets.
 
BMW is co-located with us in Munich. Their R&D people are working with our people to build the most advanced cognitive cockpit in their electric car and they want to have the most sophisticated cognitive cockpit.
 
IBM has publically cited a number of $10 billion in terms of Watson's contribution to revenue by 2023. While you don't breakout Watson specific numbes, are you at least on track in terms of progress and expectation?
 
I firmly believe that we are seeing the momentum and (we) are on track. We don't breakup the number so we can't discuss on that but what we can show you is the percentage of cognitive solutions for our total revenue. You can see cloud and cognitive solution as a percentage of our revenue, which is available publicly. You can see what was shown last year financial result and this year result. That is a significant jump from a contribution to company's revenue and that is only coming from cloud and cognitive solution.
 
For us every solution that is leveraging Watson is a cognitive solution. What we are doing now is we are making sure that we are using one standard platform. Everything is put together as one single platform and all the calls from that platform are onto Watson which are API, (We have now a library of 136 to 140 API for many different industry and different use cases) so those are creating revenue for us. The way we monetise that, we make a simple as you use service for a client in any kind of solution and they can use this on the cloud and pay for the API.
 
First of all this is a long journey for us. It is our commitment to transform the company, we are a 100 (plus) year old company. We are transforming from our dependence on traditional products to these new cognitive services. We have made great progress and we are now in the phase where we are actually capitalising the internal transformation that we did. We spent our last few years doing our internal transformation to be able to do this business. We are winning the markets now.
 
We see the tremendous progress in the market. The market definitely sees that IBM is making that shift and we have got the feedback from our clients that they feel this is the new IBM, they see a different approach by IBM, different way of dealing and a company that is truly making that shift. So I guess these are the good indicators. Stock market and client feedback are also a good indicator.
 
Specific to India Market, what are the kind of opportunity you see? You have worked with few large brands like Persistent, Tech Mahindra and you also tied up with Manipal Group.
 
I don't think we will grow without a very strong ecosystem, so ecosystem comes in different forms and shape. First of all device manufacturers are important parts of it, then we have network providers, then we have technology, we ourselves are providing multiple technologies and we also have companies that have great apps.
 
Then you have got system integrategrators, people that are actually going to help in change management and system integration. So we have our own GBS (IBM Global Business Services). But we will never be the masters of every domain. You will have likes of Tech Mahindra, Accenture, Capegemini, we work with everybody who are going to be the domain experts and going to have best relationships with certain clients.
 
For us, it's not an opportunistic thing but it is more a strategic investment thing. I am not playing with every GSI in Watson. I have chosen 5 GSIs and are partnering with, and my strategy is to be domain specific with a particular GSI.
 
For Tech Mahindra we have very strong relationship with them. It comes with a very strong manufacturing pedigree, they have a shop floor and they come with lots of M&M background. They bring engineers into their IT and they actually know the space. Our partnership with them is around industrial engineering. For example we built a solution with them, they built the industrial tool automation solution and it is tied up with the Watson IOT platform and industry solution. Together we are offering this as a solution to the customer.
 

How does the joint go-to market actually work in this kind of environment? Does Tech Mahindra fronts the whole thing to the client and pays IBM something for leveraging Watson?
 
We have two models in that. One is white labeling, basically Tech Mahindra would front it and our solutions are right behind that. And the monetisation part is clear, every time you use my platform or my capabilities a part the revenue comes to me. The other is we are the joint go to market where we co-brand, in this case we have co-branded. We have said (Tech Mahindra) industrial solutions powered by Watson.
 
It is powerful because clients like the combination of two and they understand what Tech Mahindra is bringing to the table and they know cognitive technology, IBM power is a very good to have. In this case we have a single contract.
 
Either Tech Mahindra or we sign the contract. We create a product ID and both Tech Mahindra and we can use it and we will sell the solutions from both the channels. All of IBM sellers have access to that solution and Tech Mahindra get access to IBM sellers.  We do not create product ID for everything. We choose the solution, create a product ID and we market it together through our channels. You have to be selective and also the client wants to know what is so unique about two of us coming together. Is it just a marriage of convenience or it is some sort of strategic solution that we built together.
 
So you are in no way handicapped or limited by IBM's own GBS (which in certain segments of the markets competes with the likes of Tech Mahindra or Accenture or a Capgemini)?
 
That's the reason why we are a separate unit. The purpose is to be able to allow me the flexibility to leverage all angles to grow in the market.
 
Apart from Tech Mahindra, and Persistent Technologies who are the other Indian SIs?
 
Infosys, TCS and HCL.
 
Out of the 2000 plus people in your division, how many are in India?
 
India for me is not just Indian market, the talent that we have in India, this is a global power hub.  Why are we investing so much money in India and China, is because we truly believe that we are going to get majority of the talent that we have needed for this new capability, from here. By 2020 we are going to have a shortage of a million data scientists.
 
Depending on who you ask about there are likely to be  20 million to 30 million IoT devices by 2020. When will we see the main streaming of IoT devices?
 
We are almost there. For instance on my current visit to India, I was talking to one of the large tyre manufacturers and basically we are working with them around how would you sensorize about tyre wear and tear. The cost is palatable to be able to create data value, because when you think about the value that we create for the fleets of people that operate vehciles, the feedback (such a sensorised tyre) poduct can create, the services that you create on back of that, the value outweighs the cost of putting the sensors.
 
The data is valuable enough?

It is valuable enough to start and as we start scaling the volumes, the cost of sensors are going down. I think in certain industries we are definitely at this… We were talking to a paint manufacturer in Mumbai. They wanted to put RFID on their (paint)cans.
 
We are coming to a point where people can see the use case is driving the business value.
 
Any other aspect of the business, which are relevant and we have not touched upon.
 
In particular, to Indian market, I can see that we have the fastest speed of adoption here, because there is no legacy and I see the appetite to experiment is much faster.