AI will shift the balance of power to the mega ISVs

The current market for Artificial Intelligence (AI) enabled services is not confined to lower level Machine Learning like recommending websites or products. Neither is it distorted by offerings such as Amazon’s autonomous drones that have little, if any, impact on horizontal processes. The astounding pace of change in the AI space can be gauged by looking at providers like Salesforce, who is offering standardized cloud-based services, is progressing to NLP and Computer Vision. Those capabilities include sentiment analysis, object detection and image classification.

To facilitate the discussions on how AI is impacting organizations service delivery strategies, HfS has already published the AI-enabled One Office Premier League where we explained the key technology building blocks on the AI continuum as well as assessing service provider capabilities around AI enabled services. With this report, we are trying to analyze whether with the onset of AI enabled services the market dynamics are changing.

The last few years, Robotic Process Automation (RPA) consumed the world of IT and business operations.  The RPA space is dominated by dominated by small startups like Blue Prism and AutomationAnywhere. The advent of AI is changing the dynamic with a group of Mega ISVs namely Google, IBM Watson, Microsoft, and AWS coming to the fore. With that, we are likely to see a return to the informal processes and politics of alliance management and programs. Even though AI is in a more nascent phase of market development than RPA, the value and scale of the deployments are dwarfing the RPA discussions that still mask and dominate the broader automation discourse. Regardless of the respective technology building blocks, the best way to understand the evolving AI supplier and capabilities landscape is to recognize the interplay of computing power that optimized for AI, an expansive set of algorithms, platforms enhanced for the particular use case and lastly but most importantly access to vast data sets including unstructured data. While the direction of travel is toward autonomous processes, we need people developing, improving, and refining the AI, its environment, and necessary algorithms. Thus, AI is reinforcing the necessity for organizations to drive service orchestration strategies. The use case will determine the best technical approach, but critically have to leverage commonality across their delivery backbones and develop the capabilities to route customer data through that backbone from interaction all the way to execution or conclusion. Exhibit 1 is depicting this interplay and providing examples of innovative offerings:

Exhibit 1: The evolving AI landscape

Source: HfS Research, 2017

We have to be careful not to misread the heightened market developments as that those technology building blocks can be integrated more or less as turn-key solutions. Rather it still takes a significant effort and skills to adapt those innovative offerings. Thus, we have to avoid making the same mistakes in the market communications as in the early phase of the RPA market. AI is not a turn-key solution, it is invasive, and deployments carry significant risks that need to be managed. As such we urgently need discussions on governance, in particular where solutions move toward self-learning and self-remediating engines.

A new set of partnerships and ecosystems are starting to emerge

Building on the deliberations of the notion of service orchestration and the interplay of large sets of innovative technology components, the acceleration in the build-out of AI capabilities will foster a new set of ecosystems. HfS has already analyzed the evolving Watson ecosystem in some detail. IBM has opened this ecosystem to its system integration competitors, and it is telling that organizations like TCS can articulate the capabilities more succinctly than IBM itself. Providers (read IBM’s competitors) are picking their sweet spots. Capgemini with its Watson IoT initiatives for manufacturing, or DXC Technology’s exploration of Watson APIs for improving core insurance processes. Service providers like Wipro and Hexaware have made investments in Watson Analytics for areas like customer analytics and predictive maintenance and are betting on embedding Watson in their proprietary platforms.

Watson is also a reminder of the time and effort it takes to get generic AI solutions off the ground and get it commercially viable. It was back in 2008 that Watson started to compete with Jeopardy champions. Suffice it to say newer market entrants will be able to compete with better technology, but where they will continue to struggle is around data and the breadth of API services. Having said that a new set of partnerships would accelerate the ecosystem mindset and mitigate some of those shortcomings. We have seen partnerships between Microsoft and Amazon to guarantee interoperability between Cortana and Alexa or IPsoft Amelia being integrated with multiple in-home device platforms, providing a channel for Amelia’s capabilities to millions of Amazon Echo and Google Home users. But innovative partnerships work in both directions as seen in the case of integrating Watson’s data and API services into Salesforce’ Einstein platform.

Bottom-line: IT juggernauts and data will dominate the move toward AI

Almost underneath the radar, AI projects are starting to scale up and are already dwarfing RPA deployments. The broader market has still to acknowledge the fundamental shift that is underway. However, AI projects need deep investments and lots of data. It is here where the marked difference to RPA is cutting in. Despite a lot of noise around startups and in particular FinTech companies, juggernauts like Google, AWS, Microsoft will dominate the market for precisely the reasons we have called out.

HfS is committed to doing our two cents in educating the market just as we did for RPA.  Over time we will provide more detailed analysis of the mega ISVs, innovative startups and most importantly ground-breaking ways to re-imagine the whole process rather than adding AI as a bolt-on. Stay tuned!