“When it comes to enterprise-grade AI, it has to be embedded into every aspect of the business.” – Srinivas Devulapalli, Publicis Sapient

CNME Editor Mark Forker sat down with Srinivas Devulapalli, Senior Vice President and Managing Director of Publicis Sapient Middle East, to find out more about how Publicis Sapient is providing AI solutions that are built to deliver, the issue of AI pilot purgatory, the necessity for AI to be wedded to your decision-making systems, why AI can never be viewed as just a feature – and the value humans in the loop can drive for enterprises.

Publicis Sapient is a technology powerhouse with over 30 years of experience in helping enterprises to achieve digital transformation.

In recent years, unsurprisingly it has pivoted towards providing enterprises with AI platforms and services.

Its mantra is ‘AI that’s built to deliver’.

CNME were afforded the opportunity to sit down with Srinivas Devulapalli, Senior Vice President and Managing Director of Publicis Sapient Middle East.

Devulapalli is a hugely respected and revered business leader in the world of technology, with experience building digital consulting businesses in global markets such as the UK, United States, Asia and the Middle East.

A couple of years ago, a frequently coined term among those operating in the technology space was ‘data is the new oil’.

The nascent catchphrase is ‘AI is everywhere’.

So, naturally enough, AI serves as the foundation to our conversation.

Devulapalli kickstarted our conversation by highlighting that AI can’t be seen as a gimmicky feature, and that for its impact to be truly felt across every layer of the organization then it needs to be part of the overarching infrastructure.

“There is a broad variety of thought when it comes to how AI should be used. However, when it comes to enterprise-grade AI then it needs to be embedded into every aspect of the business. It can’t be a feature. If you look at the whole lifecycle of enterprise platforms you need to use AI to improve individual productivity, that’s one gamechanger, but you also need to use AI to improve overall solution quality and scalability of solutions. Ultimately, you must embed AI to become that intelligent layer that you need,” said Devulapalli.

The head of Publicis Sapient Middle East said that there are areas where you can let AI be pretty much autonomous and make decisions to improve the quality of the solution at that stage.

“In my opinion when it comes to the enterprise, AI needs to be embedded in systems, workflows, design systems, and decision-making systems. For me it lives and breathes through the entire ecosystem, and entirety of your solutions, and that’s why we think of it as infrastructure versus just a feature,” said Devulapalli.

What has also emerged as a topic of conversation over the last number of months has been that of AI competency.

What has become evident is a clear disparity in terms of know-how when it comes to companies trying to leverage the power of AI into their operations in a bid to enhance productivity and efficiency.

Devulapalli said there needs to be internal governance when developing AI use-cases.

“We have been having multiple conversations with clients when it comes to AI, and some clients are very savvy when it comes to how they want to use AI, whilst others don’t have the same clarity. If you’re a retailer then you’re looking for specific KPIs to drive many aspects of the business, which could range from profitability to increasing revenue, if you are a grocer you could be looking at AI to optimize your entire supply chain, and what they have in common is a very specific understanding of where their business can improve if they leverage the power of AI. However, what I’ve seen quite a lot of is businesses that create a backlog of use-cases, and see which ones perform, and that’s almost like a fail-fast approach. There’s nothing wrong with it, as long as you don’t get caught up in just the use-cases, but you need some sort of governance that says what am I driving towards with these 500 AI use-cases? Where am I going? So, you need to govern what you’re trying to do and have a clear detailed plan for your AI use-cases,” said Devulapalli.

The sheer volume of AI pilots that don’t get to the production stage has been well documented, but according to Devulapalli one of the main factors for this is the challenge presented by legacy systems.

“There is a thing within the industry called ‘pilot purgatory’. There are countless examples of businesses across several industry verticals building a huge volume of pilots that ultimately never see the light of day. There’s a lot of issues in pilot purgatory, but one of the primary factors is that they are all built in isolation. There’s a huge disconnect from the legacy system. What most people do with AI pilots is create a controlled volume of data. It’s clean data, perfectly set-up, and it’s in isolation and it has sample datasets that are used to prove the pilot. However, when you try to industrialize the pilot then it becomes a challenge because you are stuck with legacy systems. The overwhelming majority of AI projects will fail if your business doesn’t modernize their legacy systems. You can’t build AI on pre-AI architecture and then hope it works,” said Devulapalli.

Publicis Sapient are on the record as saying that businesses need to view modernization as a vehicle to drive their growth strategy.

But we know that lifting and shifting legacy systems can be tedious and painful.

However, as Devulapalli points out this isn’t a case of AI vs legacy, he believes that many businesses have come to the painful conclusion that you can’t build AI on top of legacy systems.

“I think many enterprises now realize that you can’t bolt AI on top of old legacy systems, especially mainframe environments. It just doesn’t work. For example, take the financial services industry, if you’re working with AI then there are a couple of journeys that you need to go on. The first one of those is migrating to a cloud-native AI scalable architecture. When you’re making that journey you have a lot of what we call RPATHs, and there are about 7 RPATHS, which can allow you to retire some solutions, refactor some solutions and replace some solutions. There is a slew of options available on that journey to modernize your architecture, but you need to think about as end-to-end solutions that you’d actually create,” said Devulapalli.

Devulapalli reserved praise for the leadership that has been demonstrated across the Middle East in terms of what he described as ‘smart government’.

“There was a period when governments across the Middle East region were saying we want to digitize our processes and they did that, and there were no better examples of this than in the UAE and Saudi Arabia. But the push from them wasn’t just to digitize these governments services but instead, optimize them. And we know that the way to optimize it is through the implementation of AI through the entire solution stack, and that’s why they are investing so heavily in it,” said Devulapalli.

One of the ongoing debates since the beginning of the whole conversation on AI has been around the potentially corrosive impact it could have on our society.

Many analysts are suggesting that AI will be the first time an industrial revolution will result in multiple job losses for white collar workers.

Devulapalli understands some of the concerns, but he firmly believes that AI is designed to complement humans and vice versa.

He also reiterated the need for businesses to embrace the chaotic disruption of AI.

“We don’t look at it as AI vs humans. We look at it through the prism of AI with humans. Publicis Sapient has a core set of AI enterprise products that can enrich, help, and create new architecture and solutions at scale that are cloud native. At the end of the day, our solutions can fundamentally transform a business. As markets continue to become even more competitive with the advent of AI, then you can either try to defend your position, which is the old world of thinking, or try and differentiate yourself in new markets, or fundamentally disrupt your own existing business model. I believe that most of the world views AI as an opportunity to leapfrog defending and get straight to the disrupting part, because nobody is interested in the middle stage,” said Devulapalli.

Devulapalli highlighted the positive role he thinks ‘humans in the loop’ can play in this whole new AI ecosystem.

“I believe that the humans in the loop are the ones that really matter because they have the industry and domain expertise and knowledge. You can then use products to help you accelerate on that vision and then the industry experts can judge is this the right approach, is it going to hit the KPIs that we want, and you can ask it fundamental questions. For example, if I am using AI to fundamentally change citizen services, am I truly achieving what I want, which is simplicity, the eradication of friction, and ultimately getting the job done? Is AI helping me to do that, and that’s where humans in the loop can come into the fray and start adding real value,” said Devulapalli.

Devulapalli concluded a terrific discussion by highlighting some of the key factors that he believes differentiates Publicis Sapient from its market rivals.

“We’re rooted in innovation. That’s always been central to Publicis Sapient, and we’ve always been a highly innovative and disruptive company. We’ve always curated our talent pool over the decades and built them up, and I’m an example of someone that has been at Publicis Sapient for over 20 years. Due to innovation being a guiding principle of ours, it has allowed us to build our own AI products – Sapient Slingshot, Sapient Bodhi and Sapient Sustain, and navigate our ship at a much faster pace. We have a great toolbox of products and when you couple that with the very smart people that we have working for us as experts then that enables us to drive exceptional outcomes for our clients,” said Devulapalli.