Mphasis CEO Nitin Rakesh has a golden rule in business—never solve a problem after it has occurred. This is a credo that the global IT major also holds close to its digitally beating heart.
Hence, it typically asks its enterprise clients to think about their digital drop ratio, which is the number of times a customer connection or a query starts on the digital channel but eventually ends up in their call center.
Nitin Rakesh, CEO of Mphasis tells Outlook Business that this icebreaking conversation helps showcase how it can help the company develop better self-service capability by digitising every touchpoint and redesigning the customer journey using conversational design and artificial intelligence (AI). This means creating digital assistants, not just for the end consumer, but also for their agents who help those customers.
What is Mphasis’ sectoral mix, revenue wise, currently?
Mphasis was spun from Hewlett Packard (HP) and 3 per cent of our revenue is still linked to the legacy HP subcontracting business. We've been consistently transforming the revenue line over the last five years and building new engines of growth.
Currently, about half of our business comes from banking and financial services. High tech and insurance contributed 14 per cent and 10 per cent to our revenue, respectively. Launched in 2020, healthcare contributes 8 per cent to 9 per cent at a run rate of about $100 million. We also set up the logistics and transportation vertical, comprising logistics and supply chain, airlines, and transportation, which adds about 13 per cent of revenue.
About 70 per cent of our revenue is application-centric today, and around half of that is application transformation. We also help clients build new platforms, whether it's around data or using it to drive personalisation and experience transformation as well as next-generation operations.
Our service portfolio, set up four years ago, now has eight tech tribes and is used to build our capability competency constantly. These building blocks embrace the key trends enterprises need in real-time, be it cloud, data, cyber, AI or a combination of these.
After renovating these tech tribes thrice, we today have a strong focus on prepackaged solutions or deal archetypes. The idea is to create reference architectures, use IP assets, repeatable accelerators, and bundled third-party platforms and products wherever required.
How has Gen AI changed Mphasis’ business models especially when it comes to incident management?
We're in the early innings of a pretty interesting place when it comes to AI, similar to what happened with cloud and data about ten years ago. Gen AI has taken AI’s adoption mainstream, enabling almost everything in an enterprise.
Ultimately, Mphasis reflects the enterprise spending, consumption and tech patterns. So, we are obviously going to pivot very quickly too.
The Mphasis AI unit was launched in June 2023 and we already have around six ready solutions. It allows enterprises to adopt the right combination of Gen AI and conversational AI to transform ticket management experiences for their end customers and internal employees. By leveraging large Gen AI use cases, they can increase efficiency, reduce the average handling time and reduce the number of customer queries.
Similarly, when it comes to IT operations, we use pattern recognition and live data feeds from server logs to identify problems before they occur. The intent of this predictive and self-healing operation is not to solve a problem after it has occurred but to ensure there is no problem to begin with.
Eliminating these tickets versus solving tickets faster is a big business model shift for many legacy providers, whose commercial model was around being paid for ticket management. We don’t want to be paid for tickets because we want to avoid them, which is an interesting place to be in.
Similarly, in business operations like client onboarding or risk and compliance, AML can be automated significantly. For instance, sanction screening is a big cost for any financial institution, where every dollar flowing through the system must be screened. Currently, someone is researching the exceptions looking for transaction patterns that match sanctioned names. Automating a significant chunk of this process can make business operations more dynamic.
How is Mphasis using Gen AI for the transformation of legacy applications?
In areas like legacy transformation, the longest pole in the tent was always a 50-year-old application built in the 70s, with low latency, a large mainframe footprint and a million lines of code. These clunky monolithic ran critical applications and no documentation was available for their digital transformation, so enterprises were wary of touching them.
The most difficult part of transforming these legacy applications is extracting the business logic from onerous codes—many of which are dead code since they were hardly used. With Gen AI, we are trying to see whether LLMs can be trained to extract business logic from these applications. While we have already embedded that into our modernisation thesis, much work is still to be done to reshape the legacy modernisation landscape.
While much experimentation is underway with use cases being designed, what kind of incremental revenue can Mphasis gain from its various AI-driven initiatives in the future?
Almost every service line we deliver in five to seven years from now will have some element of cognitive, AI or machine learning embedded in it. To me, what's less important is what percentage of my revenue is AI-enabled. What's more important is the incremental percentage of the deal will have an element of AI embedded in it. Are we AI-enabling everything we do with the customer? This is similar to the value migration we saw over the last 25 years with offshoring and later with digital adoption.
This time, it is not just about having the right skills at the right price in the right market in the right geography, nor is size your friend. Clients are not looking at the lowest cost per unit and the value companies like us can create. So, it is about how we truly take a ‘tech-first’ approach to deliver every service through a tech-enabled platform.
We took an architecture design and engineering lead five years ago, which will give us incremental share gains. For example, IT operations is a legacy service line where we didn't have a very strong play.
So, whenever there was a large IT operation-centric deal, we would compete with many of our larger peers with thousands of people in a shared service construct. Setting up an entire unit for one deal isn’t economically competitive.
We decided to change the equation—it is not about the number of people involved in the project but how much manual effort can be eliminated. It is a more compelling case for the customer if we could reduce the number of people deployed on a $50 million global IT services helpdesk by more than half. According to early signs, this is the direction that enterprises will move towards.
I'm not a big fan of doing use cases or POCs with customers because we must have 50 pilots running on Gen AI. Our approach is to focus on a few strategic areas, like the modernisation of operations and customer experience transformation, which are likelier to see rapid adoption.
With Gen AI, IT projects will not be incumbent on the person-hours billed but on the overall speed and efficacy. Will this result in massive job losses or merely realigning the existing talent pool?
AI will not replace humans; AI will replace humans who don't use it. We will continue to ensure that every employee working on any project or program is skilled enough to use AI.
Incidentally, this is not the first time we have seen a pretty massive tech-led pivot that have impacted jobs. A study by economist David Autor found that 60% of the jobs in today’s global economy did not exist 100 years ago.
When I landed in Chicago’s O'Hare Airport 30 years ago, there was a vast United Airlines with several thousand employees. These people used to process ticket stubs from paper tickets, which was essential for paying commissions to agents. Back then, the joke was that United Airlines wasn’t a company flying planes; it was a ticket-processing business!
However, with the introduction of tickets, you can easily guess how many employees in any airline company process tickets today. Was this because of the internet and digitisation? Yes. However, United Airlines has more employees today than in 1995, except that these people are working on different tasks.
It means that new value migration levers are created with technological advancements. If people don’t upskill themselves and keep looking for a job, there won't be any.
How will working on big-ticket projects with minimal human intervention, and therefore minimal expenditure in human capital, impact your margins?
There are three aspects to the margin story. If you adopt more tech and eliminate more manual effort, theoretically, you should be able to expand your profitability.
The next question is how much profitability you will see in your balance sheet and how much you will actually need since you also have to invest in capability building. For example, if you want to operate high-tech, low-touch, low-manual IT operations, then you have to build that platform, run and maintain it, and add its latest functionalities. Some of that lift in profitability due to the elimination of manual effort will get offset because it will be ploughed back into constant capability development.
Third, this is a mature market in terms of pricing because buyers have been doing this for more than 30 years. After the first sign of competitive intensity, as you offer a tech-enabled solution, the cost per unit to the customer will go down. Also, companies with scale will anyway drop prices.
So, the competitive intensity and maturity of the buyer organisation will keep margins in check for everyone. So, is there an opportunity to go from 18 per cent EBITDA to 20 per cent EBITDA with AI and Gen AI? Absolutely. But can you go from 18 per cent to 30 per cent? I doubt it.