AI/Automation

The Way to Boost AI Adoption is through Micro Applications

Getting on the AI bandwagon is undoubtedly an aspiration for almost every organization. A prominent research firm has predicted that by 2022, as many as 40 percent of customer facing employees and government employees will consult some form of AI for decision making. While the forecast looks extremely promising, the truth on the ground is that enterprises are dealing with numerous challenges when it comes to AI adoption. Very few enterprises today can claim to be in a mature state of Artificial Intelligence (AI) adoption. In fact, several enterprises are even unsure about how to get started on their AI journey.

Understanding the AI Journey – from Deterministic to Predictive to Cognitive

Typically, the AI adoption journey goes through an automation continuum that goes from deterministic functions moving to the predictive and then to the cognitive stage. Deterministic is centered on automation of repetitive processes to improve productivity. This is the gateway to full-fledged AI adoption. As organizations mature in their AI competencies, they consider analytics-driven operations to predict failures and proactively avoid business disruptions. Finally, they move to maturity with the adoption of cognitive abilities to leverage data to derive evolved patterns for making better business decisions.

It is important for organizations to understand where they stand in their AI adoption journey as it is the starting point for them to define their AI objectives. This would help them take the next step of understanding their current capabilities, assessing their requirements, finding the right vendor and managing the change.

Every organization has a different AI maturity level and every stage requires a different approach with respect to planning, execution and management. Questions around strategy, business, technology, and employees need to be answered, for example, whether a transformative approach is better than an incremental one. What are the problems that the enterprise wants to solve using AI? Should one build or buy? What change is required in the culture of the organization to handle the change?

Finding where an organization fits in the continuum thus becomes an important step to creating the right strategy, evaluating and choosing the right solutions, strengthening governance for execution and managing organization change. This also helps to continuously discover new opportunities as the client scales up, provide the right ecosystem and implement contextualized solutions.

The AI Journey is not Free of Challenges

While most enterprises appreciate the importance of AI adoption, they hesitate due to the perceived complexity, potential disruption, and legitimate hindrances. One primary challenge, for example, is that most organizational systems and processes run in siloes with duplication of data across systems. Organizations don’t have a way to derive any compounded advantage from all their data.

Secondly, some enterprises simply lack the data, expertise and infrastructure required to build AI functionality. Besides, the pace of change in technology is high and implementation often disrupts the existing foundation of the organization. And after all this, business outcomes are often hard to map since measurement is a challenge. The lack of adequate explainability in AI decision making can also be quite disconcerting.

That being said, enterprises do see the power of AI and its potential impact on their organizations. In a recent survey titled ‘The Path to AI’ that we did at EdgeVerve with AIIA, 37 percent of respondents said they would like to implement AI in operations and customer service. Another study titled ‘Leadership in the Age of AI,’ conducted by Infosys and a Market Research firm across seven countries, asked organizations about the strategic advantages that AI is producing across industries. 45 percent picked process improvements while 40 percent said they expected productivity gains due to IT time spent on higher value work.

Mitigate the Complexities of AI Adoption with Microservices

One key way to breach some of the AI adoption challenges is to make it possible for businesses to consume AI without complexity. The answer lies in AI-powered micro-applications. These are point solutions that are plug-and-play and deliver specific business outcomes. The key is that these solutions should be easy to deploy and use and must not require specialized AI skillsets. These need to be designed as over the top (OTT) solutions that can integrate seamlessly with existing systems while removing data siloes.

In general, there are three types of data that need to be analyzed to get useful insights – the first is data from internal systems, the second is external or third-party data, and the third is the vast pool of public data. An AI-powered business app isn’t just another system of record; rather, it is a system of innovation that can pull from all these three sources to give credible insights.

This approach to AI-powered microservices is something that we follow as well. Our concept of the business app is centered on the Infosys Nia platform. All our apps are built for AI on this platform. It includes a data layer as well as data adapters to hook on to existing systems of record or other data outside the data. It also has an advanced machine learning framework, a natural language processing piece, and advanced analytics.

The complexity, however, does not percolate down to the user level since the user can access the service through a specific business app, whether for Sourcing and Procurement, Demand and Fulfilment or Finance. So, the user is not consuming a platform, but rather, just addressing a specific business problem through an app that is powered by AI. EdgeVerve’s suite of Business Applications is extensive and integrates easily with existing IT infrastructure.

A great example is a large American pharmaceuticals and consumer packaged goods manufacturer that was dealing with issues such as multiple complex systems, a big support staff, high cost of system overheads, and business processes that were not standardized. The organization implemented Nia Guided Buying that encouraged self-service and increased system adoption through a simple and intuitive user experience. The solution sits on top of existing systems and standardizes processes. The solution has helped the company increase catalogue adoption from 35 percent to 77 percent. It also reduced transactional cycle time, improved productivity and increased compliance.

As demonstrated, adopting AI-powered micro-applications can be a great way to derive the humungous benefits of AI, while mitigating some of the complexities.