Does AI have a place in Pharma? If it does, where should we start? These questions were playing on my mind a few months ago when I took the helm of the technology organization at my employer, one of the largest pharmaceuticals in India. While the rest of my Digital agenda was identified and ready for execution, AI was behind, and I had a feeling that my employer was perhaps not ready yet for AI.

Now, I think the world of artificial intelligence, but the concept is widely misunderstood, and in my view, we needed to deliberately make the decision of whether to tread into the field. The reality is, most of what passes for AI today doesn’t contain any element of real artificial intelligence.

Businesspeople mistake smart systems for AI powered systems. For example, if my HR system is voice-enabled and responds to my questions and requests, employees and HR professionals alike hail the ‘AI technology.’ In reality, the system just uses smart technology – in this case, a chat robot. AI, by definition, has to include the element of learning.

All right, so how do we proceed with AI, I wondered. We decided to start with RPA, or robotic process automation. AI is a journey, of which RPA is a good first step. You can use RPA to automate your HR processes such as reimbursement processing, or Finance processes such as payroll processing and taxation. This technology primarily uses so-called ‘chatbots,’ or simply ‘bots,’ that automatically performs manual, repetitive tasks.

This line of thinking brought us to the point where we were ready to execute an RPA project. After analyzing about six problem statements, we zeroed in on the following.

“RPA is a required step in the journey to acquire Digital capability in the form of artificial or augmented intelligence”

Employees need to travel for business purposes. This usually leads to high volumes of travel arrangements and documentation. The key challenge is the volume of invoices from the travel partners or agencies, the time and effort to process the paperwork (the invoices along with supporting documents). The repetitive nature of the task makes it eminently suited for automation. Additionally, the workforce employed in handling this task can be released for higher value-add tasks.

Now, let’s take a look at the scope of the task. For a company with $1B of annual sales, it may translate into the following order of volume of paperwork per month: 4,000 airline invoices, 6,000 hotel invoices and 2,500 taxicab invoices.

Our objectives were:

1. 75% reduction in manual data extraction in Phase 1. RPA is a technology that can be refined in order to improve the effectiveness of automated scanning and data extraction. We approached this business problem in the form of phases. In Phase 1, the bulk of the ‘easy’ data extraction would be performed.

2. 100% integration with ERP and other systems that can use the extracted data in downstream systems. All the electronic data spitted out of the RPA system had to be fully integrated with other systems.

3. Reduction of manpower requirement by 80%, in terms of FTEs. After performing an analysis of the various products in the market, we narrowed down our options to one of the following three: (a) UI Path, (b) Automation Anywhere, and (c) Blue Prism.

We then evaluated the competing proposals on the basis of a framework that consisted of the following criteria: (a) Architectural fitment, (b) Financial benefit, (c) Ease of adoption and use, (d) Time taken to realize benefit, and (e) Total cost of deployment and ownership.

Once we made our decision on the product and the partner, the next question was whether we would want to go for a full-fledged implementation, or a smaller-scale project. We decided to think small to start with – i.e., facilitate pilot projects and PoCs or proofs of concept. One of the beauties of Digital is that the philosophy of PoCs and pilots is built in; businesspeople and partners understand equally well that the outcome, or part of it, needs to come before the investment. A percentage of the pricing, in most cases, is linked to the outcome.

A word here about senior management buy-in. We followed a two-pronged approach: first, we provided annualized benefit numbers, clearly mentioning the source (e.g. $0.8M reduction of costs due to 75% lower FTE requirement). Second, we portrayed the benefits as direct, i.e. revenue increase or cost reduction. Anything else would have been indirect – e.g. increase of employee productivity. Senior managers would always ask you: so what if we increase the productivity? Can we reduce our headcount? If not, this is only an indirect benefit. And that’s a valid perspective.

As I mentioned earlier in this article, RPA is not an end in itself, at least to my company. It is a required step in the journey to acquire Digital capability in the form of artificial, or augmented, intelligence. By following a deliberate process to identify a suitable business problem, and a thoughtful approach to pick a product and a partner, we were able to take that first step in our Digital journey.