New Delhi, June 3 -- We're at a point where enterprise technology faces pressure not only to deliver faster but to rethink the meaning of transformation itself. Publicis Sapient is a global digital business transformation company that helps enterprises across industries rethink how they use technology to improve operations, customer experiences, and business models.
In a conversation with TechCircle, Rakesh Ravuri, Chief Technology Officer (CTO) and Senior Vice President (SVP) of Engineering at Publicis Sapient, discusses how Artificial Intelligence (AI) is shifting enterprise technology beyond traditional automation toward intelligent decision-making. He explains why AI-driven workflows will define competitive advantage in the coming years and how organisations must embed AI into their core operations to stay ahead. Ravuri also highlights key structural changes, platform strategies, and sector-specific challenges shaping the future of enterprise technology. Edited Excerpts:
What assumptions do you think enterprise tech still holds onto, and where do you see the industry stuck in old beliefs?
What's happening in software today is similar to the Industrial Revolution in manufacturing. Back then, manual production gave way to machines and automation, disrupting everything and forcing a rethink of how goods were made.
Software development has been a manual, human-driven process. Now, for the first time, machines, through AI, can generate software. That changes everything. Software has always been the most expensive and resource-limited part of building digital products. Companies built plans around that constraint.
AI is removing that bottleneck. But at the same time, expectations are rising. We've talked about intelligent systems for decades, but they were mostly rule-based. Now, software can actually understand user intent, without requiring users to adapt to it.
This shift means users expect software to be intuitive, smart, and outcome-focused. Companies now have to build better software, faster, and at higher quality. It's similar to what happened during Covid, digital transformation went from optional to essential. With AI, companies must now rethink how they build software: move faster, automate more, and deliver smarter systems. That's the core challenge enterprises face today.
What key structural changes have you seen in how enterprises approach technology over the past 2-3 years?
Technology started entering enterprises back in the 1990s. At that time, computers weren't common on office desks. Over the time, that changed, now it's hard to find any office without computers. That shift marked the beginning of digital transformation.
A similar shift is now happening with AI. Over the last decade, being online, having a website or an app, became essential. Businesses that didn't have a digital presence weren't considered modern. Today, the same is happening with AI. In the near future, a business that doesn't use AI across its operations will be seen as outdated.
This goes beyond software development. Companies will be expected to use AI in internal operations, hiring, training, managing people, automating workflows, and more. Not using AI will be like not using the internet in the 2010s. The expectation will be clear: how is your company using AI to improve internal efficiency and deliver value to customers?
AI will no longer be optional. It will be a core part of your technology stack. Without it, systems will appear outdated. This shift is as big as the push for automation two decades ago. Back then, automation meant moving manual tasks to computers. Now, it means embedding AI into those systems to drive decisions within workflows.
Companies will need to answer: What work is handled by AI? What's done by humans? What decisions are algorithm-driven versus human-made? Every business needs a clear strategy for how AI is integrated. Ignoring it is not an option.
Are we overusing the word "transformation"? What term fits better for what you see in large organisations?
Transformation is still essential; it hasn't gone away. But for transformation to happen, there needs to be a trigger, and those triggers change over time.
In the past, digital transformation was driven by events like Covid, which forced businesses to move online. But at Publicis Sapient, we focus on business transformation, not just digital. We call it "digital business transformation." Digital here means using technology effectively, it's not just about being online, it's about how you use computers, systems, and now, AI.
Today, the trigger for transformation is AI. It's no longer about just having internet-facing services or moving to the cloud. Earlier, in the '90s and 2000s, transformation was about automating manual tasks and centralising data. Between 2010 and 2020, it was about putting services online, using Application Programming Interfaces (APIs), adopting cloud infrastructure.
Now, the question is: are you using AI? Is your system designed to work with AI agents? That's the new architecture challenge, just like cloud-native architecture was ten years ago.
So, transformation is still happening. In fact, it's happening faster and on a bigger scale than before. Businesses now need to re-evaluate the value they offer, because many services can now be delivered by AI. This means rethinking the business model and, in turn, transforming the business itself.
The core idea stays the same: transformation is necessary. What changes is the trigger, based on technology shifts and the broader environment.
How are modern enterprises approaching platform strategy? Are they still focused on build vs. buy or centralisation vs. composability? Where is this heading?
Platforms have evolved alongside digital transformation. Initially, platforms focused on digitising data, converting paper-based or process data into digital formats. Companies invested in data platforms like databases and data warehouses to generate insights and reports.
Next came internet-based platforms. Once data and insights were in place, companies needed to expose them externally via APIs and services. This enabled collaboration through Software-as-a-Service (SaaS) models, where APIs became central to platform architecture.
After that, a new layer emerged: agent-based access. Instead of direct API calls from apps, AI agents now interact with data and services. Technologies like Model Context Protocol (MCP) servers and agent-to-agent protocols are becoming more common. To participate in these agent-driven workflows, companies need an agent layer on top of their API layer.
This brings us to the next stage: agent-native architecture. Enterprises must now create platforms that support agents internally and externally. These agents require structured access to company data, so data platforms must also evolve to provide relevant context to AI.
To do this, companies need an agentic platform, either by building their own or adopting one from a cloud provider (e.g., Azure, Google, AWS). The choice depends on scale, goals, and specific needs. Some, like financial firms, may choose to build their own for better control, compliance, and security.
A key concern with agentic platforms is control: auditing, setting guardrails, and managing what AI can access or share. Building your own platform offers more control through internal hooks and governance.
We see both approaches in practice. We assess client needs and recommend building or adopting based on their context. Internally, we've built our own agentic platform and developed our products on top of it, giving us control over how agents and AI operate within it.
How far can AI and automation go in complex enterprise systems, or are we still early in the journey?
We're still early in this. We haven't even scratched the surface yet. There's a lot of discovery ahead, standards, best practices, and frameworks will emerge over time. This is the beginning of the next S-curve for enterprise automation and IT.
For enterprise IT, we're just getting started. As real implementations begin, we'll see what works: how to set guardrails, how to evaluate AI systems, and how to ensure they do what they're supposed to do.
Traditional testing won't cut it. It's based on fixed inputs and predictable outputs, A + B = C. AI doesn't work that way. If you mix salt and water, AI might say "brine," not "salt water." The responses are contextual, so we need new evaluation methods.
Quality assurance will change. Much of what QA teams do today will be automated by AI. But new challenges will arise, especially how to test AI systems and design evaluation frameworks that make sense for them.
Security is another area where standards need to catch up. It took years to define what "secure internet" meant. The same process will play out with AI. "Secure AI" is still undefined, and we'll need a mix of research, testing, and iteration to figure it out.
How does Publicis Sapient handle transformation in the energy industry, where clients have complex physical and digital systems and can't just "move fast and break things"?
There are several ways to approach this. One effective method is to go phase by phase. Transformation doesn't need to happen all at once-it can be incremental.
In the energy sector, one of the most impactful areas for AI is integration. Energy systems often rely on multiple complex providers, and automating those integrations can bring significant value. AI can also improve forecasting and prediction models, which are critical for energy operations.
Another area is interface complexity. Energy applications often have dozens of input fields, buttons, and controls. AI can help simplify these interfaces and support users in making better decisions. This is especially useful for energy traders and practitioners who interact with these systems daily.
You can also use AI to build smarter integrations - for example, when dealing with oil, commodities, or other providers. Every industry can adopt AI in stages, deciding where to begin and where more substantial changes will come later.
Which industries are moving fastest toward full business model transformation, beyond just tech upgrades?
When you ask which industries are adopting AI, the answer is: all of them. No one is holding back. Surprisingly, industries that have traditionally been slow to modernise, like insurance and healthcare, are now moving quickly.
These sectors have long relied on legacy systems, especially mainframes, which are still used for most insurance processing and many healthcare operations.
The main challenge with these systems is their complexity and the lack of available knowledge. The people who built them are no longer in the workforce, and current teams often don't fully understand how these systems work. As a result, organisations have avoided changing them.
Now, AI is changing that. One of the fastest-growing areas of work is modernising these mainframe systems. Traditionally, we didn't work with mainframes, but with AI, that's no longer a barrier. AI helps us analyse and understand these legacy systems. Once we understand them, we can modernise them, something we've always been good at.
Clients are asking us to use AI to interpret their legacy code and rebuild their systems. There are around 1.3 trillion lines of old code out there, code that hasn't been touched because no one fully understands it. AI can read and explain this code, eliminating the need for human interpretation. Once we have that explanation, we can use AI to help build modern systems to replace the old ones. That's become one of our fastest-growing services.
What's one undervalued tech idea today that will shape enterprise competitive advantage in the next 3-5 years?
What will really matter is how you're using AI in automation. We already know that companies using automation effectively are more efficient and predictable. The real competitive advantage now is how AI is being applied to automate decision-making, making it faster and more efficient. The biggest bottleneck in most companies is their slow workflows and decision-making processes.
If AI can handle well-defined tasks automatically and only escalate critical decisions for human review, it can significantly improve workflow efficiency. That means companies can deliver value to customers faster and at lower cost.
We've seen this before, companies like Amazon succeeded because they used automation and technology extensively. They weren't bound by traditional systems and were able to run more efficient operations and offer cheaper products. The same shift is coming to all industries. Companies that integrate AI into their processes will move faster and deliver more consistent results.
Most workflows today are built around handling exceptions. Human decision-making often leads to delays or errors due to lack of familiarity with rules or processes. AI, on the other hand, can make decisions based on historical data and clearly defined logic.
This means non-critical decisions can be made instantly, and the process continues without delay. Only a small percentage, maybe 5%, of decisions that actually need human input will be flagged. The other 95% can be fully automated. Companies that use AI effectively in their internal operations will have a clear edge going forward.
Published by HT Digital Content Services with permission from TechCircle.