Data Analytics

AI- Going Beyond Labor Substitution to Data-Driven Experiences

Many enterprises fail to capitalize on the potential that artificial intelligence and automation have to offer by equating it with labor substitution. For instance, I was recently interacting with the client team of a large bank. Their target was to reduce costs of internal and customer transactions by implementing 500-600 Robotic Process Automation (RPA) bots in a single year. They had already tried doing this with a toolset they had invested in, but were unable to meet the target. A few questions on overall process design revealed that they had not really thought through their plan. For instance, they had identified 50 use cases for immediate automation since these had a large number of people associated with them. However, the company did not know how to systematically identify new problem opportunities in the enterprise. I encountered a similar situation in a large manufacturer that was looking at IT outsourcing. They were keen to learn how artificial intelligence (AI) could help them automate work drivers and outsource the rest of the work. A retail customer, on the other hand, had already implemented RPA but needed assistance to leverage AI to respond to evolving consumer expectations.

Consistently engaging in problem finding can be a struggle for enterprises, especially if they do not refer to the data existing within the enterprise. Many of today's enterprises find themselves overwhelmed by data. Unlocking the right data and making it available across the board gives users the opportunity to engage with stakeholders in a more meaningful way than ever before.

Adopting the right approach to AI and automation is crucial in data-driven enterprises. Here are a few pointers that should help you set off on the right course.

Process discovery and design

In the automation journey, the obviously manual processes are the first to be identified. However, rather than a one-time activity, it is necessary to continuously identify automatable processes. For instance, the bank I wrote about earlier was shown how to deploy a 'problem finder' probe by Infosys. The probe unearthed insights on how end-users worked with the current enterprise resources. This enabled the enterprise to identify areas of redundancy and optimization. Sometimes, analytics from these probes indicate that the enterprise needs to digitize their operations more comprehensively, and capture data from their customer interactions, supply chain, equipment, and internal processes to make the right decisions.

In process discovery and design, enterprises have the opportunity to completely reimagine a customer engagement - right from the way data is generated, collected, organized and acted upon. In a retail scenario, it would require knowledge about the brand a customer likes, when the product needs to be replenished, ordering it for the customer in the right amount and variant through a virtual assistant, and placing the order with the right retailer.

Operational execution in automation process design

Enterprises must plan on how to respond to scenarios where robots repeatedly fail. Robots may fail due to changing business processes, and redundancy of the past method of resolving transactions. A scalable method of recording changes in business processes based on actual data, and maintaining a digitized reference-able knowledge hub of the initial process as well as changes, can prevent this.

In industries where enterprises rely on a standardized set of data for decision-making, introducing fresh orthogonal data (data that can be used without considering its effects on other program functions) to supplement data sets already in use can change the basis of a product design and experience. This is how, for example, a large CPG enterprise can identify anomalies in the customer sentiment towards their product, and a positive sentiment towards the competition. This opens up an opportunity for possible change in the customer's product experience.

The senior executives of the large bank with whom I was interacting were impressed by the scalable and repeatable ways of introducing data gathering probes into the various sources of truth, and how these could discover automation opportunities across business and IT operations.

Driving next-generation operating models

In the case of IT outsourcing that the manufacturing client was considering, an AI system of systems could completely reimagine their current model of executing IT programs. The erstwhile 'people only' model can be replaced by a 'people + software' model. So what does this mean? The new operating model could enable 'robot personas' to do the work, and augment it with human personas wherever needed. For instance, the role of an Oracle database administrator in an IT outsourcing deal can be fulfilled by a 'robot DBA' avatar. There can be a similar robot meeting the requirements in an infrastructure scenario or as a 'procurement specialist' in a sourcing-and-procurement business process function.

In conclusion, organizations preparing to automate need to adopt a sustainable process design. Knowledge gained from their current data-driven experience is key to building an automation-based foundation. This knowledge is the fabric of the enterprise. It constantly evolves and learns from past decisions so that evidence-based decision-making gets better with every passing day.