Why AI Initiatives Must Start With the Business Case
— Santhosh Subramanian, Conduent

Santhosh Subramanian has spent roughly 24 years in the IT industry, beginning his career as a software developer before progressing through architecture, technology programme management, and portfolio leadership. A pivotal moment came when Capgemini sponsored his TOGAF certification, and he took on his first profit-and-loss responsibility, a shift that moved his focus from pure engineering towards business outcomes. 

Today, at Conduent, Santhosh serves as Director of Transformation within the Office of the CIO, reporting to the CEO’s office. He oversees automation, software development lifecycle strategy, demand and supply management, and financials for an IT portfolio of roughly $100–150 million supporting a $2 billion commercial business, while also acting as chief of staff and advisor to the CIO, translating technical execution into board-level strategic value.

From Technical Architect to Business Leader: What Was Your Turning Point?

Santhosh traces his shift from engineering to business leadership back to a single certification and his first experience owning a budget.

“I started my career in the early 2000s as a software engineer, working purely on programming languages and architecture for about ten years. The defining moment for me came when I became a technology programme manager and was given a budget for the first time, at Capgemini, which sponsored me to become a TOGAF-certified architect. That certification really changed how I looked at things. 

I realised that the best technical architecture, if it doesn’t translate into profit and loss or business impact, doesn’t actually matter. The true skill was never just about writing code; it was about managing delivery, managing P&L, and managing business and client expectations. That pivot took me from leading small engineering teams to leading well over a hundred people across multiple geographies.”

Why Should Technology Never Lead a Transformation Strategy?

Reflecting on decades of technology cycles, Santhosh explains why organisational change, not the tool itself, has to come first.

“Technology change has always been happening, from the internet and Y2K to mobile, cloud computing, and now AI. But technology change can never happen in silos; you have to ask what the business is trying to achieve and what value you’re delivering, because technology is an enabler, not the starting point. You should never start with technology. Keep the business outcome in mind, and then work out how technology aligns to it.

We always say people, process, technology, not technology, process, people. You start with people, then the business process, and only then the technology.”

What’s the Biggest Misconception Leaders Have About Adopting AI?

Santhosh argues that chasing quick wins with AI, rather than starting from a genuine business case, is where most organisations go wrong.

“If you start with technology, you’re not on the right path, even if you don’t realise it immediately. People go into a meeting, see a flashy product, and sign up for it without understanding its full capabilities or drawbacks. In my organisation, we don’t deploy AI just to harvest low-hanging fruit and book a quick cost saving. We start with a business case: what is our end-to-end process actually worth?

We process around five billion documents annually across Conduent, and the question is never simply how do we cut costs, but how do we scale, and where can AI genuinely add value across that workflow?”

Why Keep Humans in the Loop When AI Can Automate the Work?

Santhosh describes the reasoning behind Conduent’s decision to redeploy people rather than reduce headcount as automation expands.

“That decision came from corporate leadership and the business leaders together. The idea was never about low-hanging fruit; if you integrate AI into a workflow and immediately cut your labour, and then a new document type comes in that the AI hasn’t been trained on, you suffer. We look at AI from both a growth standpoint and an efficiency standpoint, building scalable, repeatable systems so that when we onboard a new client, we can reuse a large part of that workflow out of the box.

Yes, we want to reduce manual, repetitive work, but we’re pivoting from being a purely services-based, operations-based company towards something closer to a product company.”

How Will Agentic AI Reshape the Product Lifecycle?

Looking three to five years ahead, Santhosh sets out how specialised AI agents could take over each stage of the development lifecycle.

“Today, a traditional software development lifecycle has five to ten manual steps, from requirements gathering through development, testing, deployment, and post-support. AI agents are going to change that.

You could have a requirements agent that captures a meeting and extracts the functional and non-functional requirements, another agent that turns that into a proper business requirements document, and further agents for build, deployment, and testing. It won’t matter whether you’re building a bespoke solution for a client or a product for a broader market; AI is going to be embedded in every step of the workflow.”

What Leadership Principle Has Remained Constant Throughout Your Career?

Santhosh describes how he mentors technical specialists by deliberately exposing them to perspectives beyond their own discipline.

“For people who look to me for mentorship, I talk about a forced perspective shift. I might be working with a brilliant architect who is excellent at his craft, but on his own he might only be seeing ten or fifteen per cent of the whole picture.

You have to bring people into other conversations, with finance, with the business, with sales and marketing, so they hear other perspectives. I never force it on anyone; if someone finds it isn’t for them after a couple of meetings, I let it go. But for me, that holistic exposure, hearing from finance, from the client partner, from go-to-market, is what builds a 360-degree leader, and it has been essential to my own growth.”

What’s the One Question Every Leader Should Ask Before Launching an AI Initiative?

Santhosh closes with the single question he believes should precede any new AI investment.

“Before launching the next AI initiative, a leader has to ask: what is the current business problem I have that hasn’t already been solved by every technology investment I’ve made so far, and what value will this new initiative actually bring? It always comes back to business value. In our industry, for example, we’ve moved automation from thirty to forty per cent with OCR, up to sixty or seventy per cent with LLMs and deep learning, yet the industry is still roughly eighty per cent paper-based.

So before any new initiative, you ask what the business case is, what the timeline is, and whether it can be rebuilt as something reusable across multiple clients. That’s what determines whether you go ahead.”

Join the Conversation: The Ortus Club’s Executive Network

As Santhosh’s experience shows, resisting the pull of flashy technology in favour of disciplined, business-case-led transformation is not a natural instinct; it is a skill built through exposure to other perspectives. Leaders navigating agentic AI, cross-functional delivery, and large-scale automation increasingly rely on peer dialogue to pressure-test their thinking before committing budget and headcount.

His approach to mentorship, deliberately pulling technical specialists into business, finance, and client conversations, reflects a broader truth: the leaders best equipped for what comes next are rarely the most siloed experts, but the most cross-functionally fluent ones.

At The Ortus Club, we host curated executive roundtables that bring together senior leaders facing these exact challenges. Join our network to exchange insights on agentic AI, business-case-led transformation, and building teams that scale without losing the human element.

Frequently Asked Questions

Q: What does “people, process, technology” mean in practice?

A: It is a sequencing principle: transformation should start by understanding the people and business process involved, with technology brought in afterwards as the enabler, rather than starting with a tool and working backwards.

Q: Why does Conduent keep humans in the loop instead of reducing headcount through AI?

A: Redeploying people, rather than cutting them, protects against the risk of AI failing on new or untrained scenarios, and allows the organisation to build scalable, reusable systems that support growth across new clients rather than short-term cost savings alone.

Q: What is agentic AI, and how does it apply to the software development lifecycle?

A: Agentic AI refers to specialised AI agents handling discrete steps of a process, such as requirements capture, documentation, build, deployment, and testing, that can eventually hand off work to one another across the wider product lifecycle.

Q: What is a “forced perspective shift” in leadership mentorship?

A: It is the practice of deliberately exposing technical specialists to conversations outside their discipline, such as finance, sales, and client discussions, so they build a broader, more business-aware view of their work.

Q: What question should leaders ask before starting a new AI initiative?

A: What business problem remains unsolved despite existing technology investments, and what tangible value will the new initiative bring, including its timeline and potential to be reused across multiple clients or business units?