Anand Haridass, Chief Engineer, Cognitive Systems, IBM and Viswanath Ramaswamy, Director, Systems, IBM India/ South Asia, in a joint conversation with Amit Singh, highlighted that traditional IT channels have a long way to go in the AI space and to gather skills in AI framework, processes, and algorithms. At the same time, they recommended channels to build synergy with AI start-ups to leverage their deep skills in niche AI solutions. They underlined that most of the AI adoption in India is going to take place this way
Despite big wigs investing heavily in the AI technology, so far the AI space in India has been led by the large set of start-ups. Do you see this as an opportunity?
Anand Haridass: It’s a mixed bag when it comes to enterprise’s preference of the vendors for AI. The large enterprises are quite protective of their data and are reluctant to work with small players.
In many cases, BFSI customers work with smaller start-ups in 1-2 small projects to test the solution but they tend to share a limited amount of data.
Moreover, most of the AI start-ups have niche solutions specific for certain use-cases and for certain segments.
Hence, in many of the cases, small start-up companies offering niche AI solutions work with large technology providers like IBM, TCS or Wipro to capture the large enterprise customers, specifically in telecom and BFSI.
In a few cases, these start-ups are acquired by the large technology providers to build and offer a comprehensive solution.
So while a large amount of innovation and niche solutions are coming from the large set of AI start-ups, they still work in tandem with the large solution providers to offer a complete solution.
In fact, we are seeing a significant amount of synergy between start-ups and solution providers/vendors. Solution providers are leveraging on the deep skills of start-ups in niche areas like medical imaging, medical diagnosis, logistics, and banking processes. Solution providers are complementing them with their extensive reach and system integration skills to offer an end-to-end solution on AI.
In fact, most of the AI adoption in India is going to take place this way.
AI requires deep skills in process optimization and algorithms. What is the level of preparedness of IT channel partners in AI?
Viswanath Ramaswamy: Unlike any other traditional software or workloads, AI requires a slightly different perspective in terms of designing a solution to differentiate or improve on the traditional processes. Hardware is just one component, understanding the existing process is quite critical in developing the framework. Only then we will be able to make the process more efficient, accurate and repeatable, which require algorithms and applications. Hence, there are three important components: hardware, framework, and algorithm. And no AI system will be successful if any of the three components is not efficient enough.
The traditional IT partners are yet to skill themselves on AI frameworks, understanding of processes, and the algorithms. Another challenge is to amend the algorithms to better train the hardware for optimum results. Hence most of the traditional IT partners have a long way to go in the AI space.
On the other hand, we are seeing a distinct ecosystem of AI solution providers emerging, which have deep skills in processes and algorithm development for specific segments. Most of them are start-ups which are developing niche AI solutions for segments including BFSI, telecom, healthcare and process re-engineering.
We are complementing them with our best-of-the-class Power AI systems to reduce the time required to train the machine. Power AI systems also relieve them from the task of going and researching the open source framework all over again; it offers them a software AI package which they can connect to any framework and it works as underlined foundation so that all the repositories, libraries and compilers are readily available to them. Hence, it enables them to write efficient algorithms much faster.
Currently, we have 15-18 AI partners which we are scaling at a fast pace.
What is the role you see for traditional system integrators in the AI space?
VR: We are building a lot of COE labs to help system integrators build AI applications for their set of customers. We are also working with many research centers where we are complementing their research with our AI experts to build solutions on our hardware.
Since AI is a new area which requires a different approach and deep skills in process, framework, and algorithm, we are helping system integrators build the new approach and deploy solutions. In the meantime, they must approach the AI start-ups to jointly build and offer the AI solutions to their customers. Since the new generation of AI partners is not skilled enough in hardware, system integrators can complement them with their core hardware and solution designing skills.
Gradually over a period of time, they can build competencies and skills in the AI domain and start building AI solutions on their own.
Most of the analysts identify lack of skills and trained resources as a major challenge hampering the growth of AI market in India. How do we plan to tackle this challenge?
VR: Almost 1.5 years back we identified that there will be an upsurge in the AI space, which will lead to a deficit of skills in the market. Hence, last year we initiated a unique Power AI University program where we are offering hands-on dedicated industry training to the graduates followed by IBM certification in order to meet the demand for AI-trained resources. We have already tied-up with four large universities in India.