Artificial intelligence has progressed to a point where it can be used to solve more than just a set of problems. But even as barriers for its adoption are lowered, thanks to the availability of ML tools and technologies, enterprises are still grappling with applying AI applications for critical tasks. Even though AI has received enormous attention in India, organisations are still looking to understand its true potential and how to realise the return on investment from the technology. Also, another big question the C-suite wishes to understand is the maturity level of the technology in the next two years.
The cost-prohibitive nature of technology and vendor lock-in that comes in with cloud companies such as AWS, Google and Microsoft, among others, who offer machine-learning-as-a-service (MLaaS), also acts as a deterrent for small and medium businesses. Many startup founders offering industry-specific vertical solutions have often claimed about clients griping about the scalability, since big vendors charge per API and few even worry about their data leaving the premises. Despite the boom around ML in India, it is an expensive proposition and hard to implement, since there is also a dearth of ML engineers.
Today, CIOs are grappling with the same question: With everyone deploying AI, are we being left behind? Well, the enterprise AI market is getting highly competitive with startups and boutique vendors offering solutions targeted at specific industries to perform niche tasks which big tech vendors like Google, Amazon, Microsoft and IBM are not able to deliver.
For example, Senseforth’s enterprise AI bots, built on proprietary technology directly compete with big tech vendors like AWS and Microsoft. This Silicon Valley and Bengaluru-based startup has the highest number of active bots in deployment — 30 — across a range of industries.
According to a research report, by the year 2019, AI and ML startups are projected to drive AI capabilities forward, disrupting the market with niche business solutions. And large AI vendors are adjusting their pricing strategies to compete with smaller competitors that are threatening to upend the Enterprise AI market.
Factors To Keep In Mind Before Onboarding Enterprise AI Vendors
The global enterprise AI market dominated by SAS, Microsoft, AWS, Intel, HPE and Wipro among others is estimated to reach $6 billion by 2023. According to a report by Market Research Future, India is advanced in software development and solutions and start-ups are also building their base in AI technology providing services and solutions to different industries.
Despite the vendor boom, organisations can be overwhelmed by the number of vendors, big and small and AI solutions on the market. For example, an IDC whitepaper about AI adoption in India discusses how few vendors may be able provide the appropriate AI software solutions but may lack capabilities to impart training to its client’s employees. Another major roadblock to adoption is a lack of strong AI advisory from top consulting firms on implementation and its capabilities.
Before Organisations Go Shopping For Vendors, Here Are A Few In-house Competencies They Should Beef Up
- Data Collection: For ML to work, organisations need to large quantity of accurate and clean data and most organisations cite combing siloed data as the biggest hurdle. For example, Manipal Hospitals which leverages IBM Watson for Oncology reportedly draws data from more than 300 medical journals, 200 textbooks, and nearly 15 million pages of text.
- Identifying The Right Use Cases For AI: For example, IT processes and sales and marketing functions are some of the prominent functions that are being automated and are improving organisational efficiency.
- Right Leadership: Organisations who are willing to explore AI implementation need senior management buy-in for enterprise AI adoption. While AI is seen as a magic bullet for solving critical problems, business leaders should also ask the right questions on how to align it with business objectives to reap tangible benefits.
- Building In-House Talent: For organisations that are still mulling the implementation of AI, talent issues are also a big priority and the shortage of skilled professionals to run day-to-day operations also act as a barrier in adoption.
Here Are A Few Points Tech Buyers Should Consider
- Vertical Generic/Cross Domain Solution: Now startups usually offer vertical solutions while big tech enterprises offer cross domain or generic solutions. Most organisations do not like dealing with multiple vendors and would either opt for a generic solution while a few, in a bid to avoid vendor lock-in opt for startups that offer industry-specific enterprise AI solutions. Choosing the right vendor or solution can help organisations avoid huge costs and reap tangible benefits. For example, Texas-based MD Anderson, the cancer research institute grabbed headlines last year for severing the contract with IBM Watson as it failed to realise tangible benefits.
- Alignment With Business Objectives: This is the most important criteria for tech buyers before onboarding the AI vendor. To avoid risk of failure, organisations should identify the right use case and start by deploying it on small scale projects. Top management should also define the key metrics they wish to achieve from the AI solution. Before the AI implementation, senior management should also put a framework for data collection and also take into consideration governance and regulatory implications.
- Strong AI Advisory: Most organisations also look for strong AI advisory during and after implementation as they need walkthroughs through each and every deployment. Also, organisations want support on training their staff on handling day-to-day functionalities as they may lack data scientists or ML professionals to handle the deployment.
- Cost Competitiveness: While on one hand AI solutions promise real benefits, we often hear C-suite talking about the cost of solution being too high. Startups which offer competitive pricing solutions often score over their enterprise counterparts in this area.
Try deep learning using MATLAB