A CTO's view on responsible, human-centred automation for enterprises

As enterprises race to embed artificial intelligence across their operations, Birlasoft has chosen to first transform itself before transforming its clients. Under the leadership of  Ganesan Karuppanaicker, CTO, Birlasoft. The company has undertaken a bold AI-first reinvention, rewiring its core processes, building agentic AI platforms, and cultivating enterprise-wide AI literacy. In this conversation, Ganesan shares how Birlasoft’s transformation began with its internal Optimus and Eagle Insight initiatives, evolved into platforms like COGITO, LINX.AI, and SIGMA, and now aims to scale responsible, autonomous, and human-centered AI across every function.

Can you elaborate on the concept of “rewiring” Birlasoft. what that entails?

The Rewiring Birlasoft  initiative builds on our internal AI foundation. We identified five key functions; talent supply chain, HR, finance, procurement, and delivery excellence, and reimagined them with an AI-first mindset. Unlike traditional process studies, we conduct  work and time studies  to understand how AI can take the first call in a process, with humans stepping in only when necessary. The principle is simple: let intelligent agents handle what they can and bring humans into the loop only when needed. This balance between automation and human intervention defines our rewiring approach.

Birlasoft has also built specific AI orchestration platforms. Can you tell us more about those?

We have developed three  agentic orchestration platforms  to accelerate app engineering and intelligent automation. The first is COGITO, a platform designed for application engineering from requirement elicitation to deployment. It supports the entire engineering lifecycle, including product owners, scrum masters, architects, developers, and testers, through specialized large language models (LLMs). The second is LINX.AI, which focuses on the testing lifecycle. It sits atop automation, performance, and security testing tools and manages the full test process through AI. The third platform, SIGMA, is a comprehensive IT operations management system developed in partnership with ServiceNow and Aisera, featuring an AI orchestration layer that integrates across multiple tools. Together, COGITO, LINX.AI, and SIGMA form the backbone for building and deploying generative AI-based applications faster and more efficiently, both internally and for clients.

As AI becomes increasingly autonomous, how do you balance innovation with cost efficiency?

Balancing innovation and cost is crucial, and we take a three-pronged approach. First, we leverage open-source technologies and cloud hyperscalers, which give us the flexibility to scale quickly and cost-effectively. Second, we focus on productivity-driven cost savings, using generative AI to improve efficiency and repurposing those savings into innovation investments. Third, we follow an ROI-first evaluation model, where every use case or pilot undergoes a thorough return-on-investment assessment. We prioritize initiatives that can deliver measurable returns within 12 to 18 months, ensuring that innovations are self-funding and cash-flow positive. This approach not only keeps costs in check but also allows us to help customers follow a similar sustainable innovation path.

Birlasoft emphasizes responsible AI. How does your governance framework work in practice?

Our responsible AI governance framework evolved through experience. Our first generative AI use case, called B-Hive, was launched in early 2023 to manage HR-related queries for 12,500 employees. It was designed to handle around 2,400 policy and HR shared service requests per month using OpenAI’s GPT models.

During this pilot, we identified several governance priorities. We implemented hallucination control mechanisms such as RAG (Retrieval-Augmented Generation) and graph-based validations to ensure accuracy. We also established an AI Risk Advisory Board focused on security, explainability, and transparency, along with a Change Advisory Board (CAB) for AI models  that assesses AI-related risks, such as switching from one model (e.g., GPT-3.5) to another (e.g., GPT-4).

To ensure auditability and control, we began with a restricted rollout to 300 users, tracking every prompt and response to evaluate relevance, hallucination frequency, and user acceptance. Continuous reinforcement learning from feedback  further refined performance. We even included customers in our AI Risk Advisory Board, since platforms like COGITO and LINX.AI are customer-facing. This joint governance approach fosters transparency, trust, and shared accountability.

You have achieved remarkable AI literacy across the organization. How have you translated that into tangible skill-building, especially for non-technical employees?

We launched the  AI for All  initiative to make every employee, from HR and finance to legal and procurement, AI-literate. Birlasoft was one of the early adopters of Microsoft 365 Copilot  in India. We started with 1,000 licenses for non-IT teams and partnered with Microsoft to deliver targeted training on prompt engineering and function-specific prompt libraries. For instance, HR teams received HR-centric prompts, while finance teams had their own sets.

We also organize monthly events, GenAI Day  and Copilot Excitement Day to encourage experimentation, gamify adoption, and recognize top “prompt engineers.” After six months, we began tracking “assisted hours,” a metric designed to measure productivity gains through AI. Once teams achieved proficiency, we expanded to another 1,000 licenses and introduced Copilot for Chat to all 12,500 employees, along with Copilot Studio, allowing them to build their own AI agents. This strategy not only upskilled employees but also made AI adoption measurable and scalable across the organization.

There has been much discussion about the proposed USD 100,000 work visa fee in the U.S. How could such a policy shift impact mid-tier IT firms like Birlasoft?

It is an evolving situation, and every day brings more clarity. What’s certain, however, is that the global landscape is shifting rapidly. While such policies pose challenges, they also open new opportunities. India today is the fourth-largest economy, on its way to becoming the third within a few years. With over 4,500 AI startups, 30 percent of which were founded in just the last three years, and a rapidly expanding data center ecosystem of over 200 facilities, the momentum is strong.

The government’s AI Mission, valued at over Rs 10,000 crore, and nationwide AI skilling initiatives have created a ready talent pool. Even our HR and finance teams now use AI tools daily. Rather than seeing this as a setback, I view it as an opportunity for India to retain and utilize its talent domestically, delivering global innovation from here. The potential to fix legacy tech debt and rewire global business processes is immense.

How do we absorb this influx of skilled professionals staying back in India or returning from abroad?

That’s a critical question, and I believe the definition of talent itself is changing. We no longer need traditional programmers, we need solution developers who can leverage AI assistants and co-pilots to deliver outcomes faster. The new metric is multiplier productivity, where we now talk about “3x,” “5x,” or “10x developers,” depending on how effectively they collaborate with AI tools. Those who master this synergy between human creativity and AI augmentation will find enormous opportunities in the evolving landscape.

To conclude, what does the next phase of AI at Birlasoft look like?

The next phase is about scaling agentic AI across every function, making our platforms more autonomous, explainable, and human-centered. Our focus will remain on responsible innovation, workforce empowerment, and measurable business outcomes. We believe in transforming ourselves first, through programs like Optimus, platforms like Eagle Insight, and agentic systems like COGITO, LINX.AI, and SIGMA, before helping our clients do the same. That’s the Birlasoft philosophy: AI-led, data-driven, and human-empowered transformation.