In the rapidly evolving world of technology, the roles of leaders like Rakesh Ravuri, Chief Technology Officer & Senior Vice President – Engineering, Publicis Sapient, have never been more critical. As organisations grapple with swift technological shifts, the post-pandemic era has seen transformations that once took a decade now crammed into mere years. In this candid conversation, Ravuri shares insights into the disruptive intersection of AI, digital transformation, legacy modernisation, and the changing face of business and skills in the enterprise technology landscape.
The pace of technological change seems to have accelerated since COVID-19, with transformations happening in years instead of decades. As CTO, what excites you most about the intersection of technology, consulting, and business outcomes? How has your role evolved in shaping this convergence?
Technology is, by definition, always changing. However, what we’re seeing now goes beyond gradual improvement, it’s a major disruption, similar to an industrial revolution for IT. Traditionally, IT has been “knowledge work” handled mostly by humans, but now we are at a point where many tasks will be performed by machines. Just as manufacturing shifted from manual labor to automation, software development is undergoing a similar revolution.
As CTO, my role is to constantly reassess how software is developed, rethinking processes that can be offloaded to AI and agents, and identifying where human involvement adds value. What excites me most is that we’re now able to focus on more futuristic projects. For example, features that once seemed out of reach, such as seamless voice interfaces, are now common. Today, even small applications can leverage solutions that once required the technological might of companies like Google. The possibilities have become enormous, and my focus is on optimising what we can offload to technology so that we can tackle greater challenges and rethink the software development process altogether.
Post-pandemic, with AI and generative technologies like LLMs evolving rapidly, how do you balance experimentation with enterprise-grade deployment, particularly in regulated industries like finance and healthcare? Can you share real-world use cases?
Surprisingly, regulated industries are among the first to adopt these new technologies. Sectors like insurance, finance, and healthcare have long relied on legacy systems, often built decades ago by teams who have since moved on. These organisations were wary of updating their systems because of their complexity and compliance risks. AI is changing this. We’re now able to use AI to understand and modernise these legacy environments safely and efficiently.
For example, our fastest-growing business is converting old applications, like those written in COBOL, into modern platforms using AI, with significant success in healthcare and banking.
Has change management been a challenge in these modernisation projects?
Change management actually comes into play later in the process. The initial hurdle was simply understanding complex legacy systems, something AI now enables without disrupting existing operations. Once AI interprets and helps convert legacy code, organisations can run new and old systems side by side to validate accuracy.
Change management becomes relevant when existing human tasks are automated, such as software testing. Here, roles evolve, from being the producer of test cases to a reviewer and mentor for AI-generated test scripts. The adaptation process involves learning to provide effective feedback and prompt engineering, which essentially means learning to communicate with AI agents efficiently.
Internally, we train our teams in these new skills: out of 10,000 people, thousands have already undergone foundational AI training, starting with prompt engineering and expanding into integrating AI into broader workflows.
Have you developed any in-house LLMs?
Developing an in-house LLM is hugely resource-intensive, only a handful of global companies can do it given the scale required. Our focus is on applied engineering: we creatively combine existing open-source and commercial LLM tools to solve real business problems. While we don’t build foundational models ourselves, we do fine-tune them for clients’ needs, such as customising models for brand-specific image generation.
Can you give examples of how these AI technologies are being used in internal operations and client projects?
We use AI both for our clients and internally. One example is the modernisation of legacy codebases: using a chain of AI agents, we reverse-engineer and rebuild old applications for industries like healthcare, banking, and retail.
In the media space, Publicis (our parent company) employs AI for marketing automation. For instance, audience segmentation, which previously took weeks, now happens in near real-time using AI-trained models and vast data pools. The system interprets campaign briefs, targets the right audience, analyses their behaviours, and auto-selects the most effective creative assets. What began as an internal platform is now being offered B2B to Fortune 10 clients.
We also built a comprehensive software platform supporting our SPEED (Strategy, Product, Experience, Engineering, Data) methodology, ensuring that every aspect of a solution is connected and collaborative. Rather than just integrating systems, we focus on system collaboration, where intelligent agents actively work together to deliver business outcomes.
How are you leveraging cloud-native architectures and modern platforms to improve time-to-market and manage long-term operational costs for clients?
Cloud-native architectures are now essential. The COVID-19 pandemic made it clear that cloud was a necessity, organisations that weren’t cloud-native struggled to support remote access and to handle sudden spikes in usage. For one Canadian retail client, for example, online traffic increased tenfold in a single day due to shifting government regulations, a challenge only cloud-native platforms could address efficiently.
With AI, the scalability demands are even greater, as AI agents interact with systems much faster than humans ever could. Cloud-native design has become the foundational requirement for digital, and AI, operations to scale flexibly and cost-effectively.
You often mention “business transformation” beyond mere digital transformation. How does your role enable clients to reimagine business models in this new era?
AI transforms not just the tools we use, but how businesses operate and create value. In the past, digital initiatives focused on getting customers to a website or platform. Now, the goal is to ensure that AI agents, not just human users, can understand and recommend your offerings.
This means businesses must structure their content and data so AI can access and promote it. For instance, marketing tactics like keyword stuffing are becoming outdated. Instead, we focus on making our services or products easily interpretable by AI, which in turn guides users or recommends resources to them.
Interaction models have also changed: instead of just presenting information, publications and businesses need to anticipate queries and structure content for conversational agents. Every industry, from retail to media to automotive, is grappling with dramatic changes in business model, with value creation now often routed through digital agents and AI systems.
As ESG goals become more tech-driven, what roles do technologies like AI, IoT, and data analytics play in helping organisations build sustainable and responsible digital systems?
The primary challenge with ESG has historically been transparency, data was hard to verify and integrate. AI and intelligent agents can help both companies and consumers access, understand, and validate sustainability claims. Businesses will be expected to expose relevant data in machine-readable formats, enabling AI agents to analyse sustainability practices and inform customer choices.
While concerns around the energy consumption of AI are valid, the trade-off is a more knowledgeable, empowered consumer base and a new level of transparency and accountability in ESG reporting.
Do you observe a skills gap among customers in their digital transformation journeys? How are these skill requirements evolving?
Rather than a skills gap, I see a need for new skills. Everyone, from writers to product managers, now needs at least a foundational understanding of how to use and guide AI. It’s much like moving from manual to automated processes in manufacturing: the core work remains, but new tools demand new expertise.
At Publicis Sapient, we define four levels of competency: using AI, programming AI, creating reusable AI-based agents, and building full AI products. Around 5,000 employees are actively being trained at various levels, using materials and tools enhanced by AI itself to create a dynamic, interactive learning environment. Traditional training methods are also being transformed with AI-driven personalised education.