Posted by Fulcrum Digital’s Founder and Chairman Rajesh Sinha on how AI is shaping enterprise IT decisions, why core systems still matter, and how a hybrid services-and-platform model is taking shape.
AI adoption is pushing enterprises to revisit long-standing assumptions about how technology should be built and consumed. While core systems continue to run critical business operations, much of today’s innovation is happening around them, through AI agents and automation layered on top. For Fulcrum Digital, this has reinforced a hybrid approach that blends engineering services with platforms and ready-to-deploy AI capabilities.
In this interview with TechHerald, Rajesh Sinhna, Founder and Chairman of Fulcrum Digital, discusses how enterprise expectations are evolving, how the buy-versus-build debate is changing in practice, and how Fulcrum is adapting its business model, talent plans, and growth priorities as AI moves from experimentation to execution.
Edited excerpts…
Q1. Fulcrum operates largely in the IT engineering and digital transformation space. With the rapid adoption of AI, machine learning (ML) and newer technologies, how has the traditional IT services model changed in your view?
Rajesh Sinha: If you look at Fulcrum’s journey, innovation has always been part of our DNA. Over the years, we have evolved through multiple phases. In the early stages, we worked on web technologies, portals, content management and social analytics.
Later, we moved into building platforms, particularly in education, and launched SaaS products like Culinary Digital. We also developed low-code and no-code platforms while continuing to deliver engineering services. This hybrid approach has been consistent for us. In 2023, we built the Ryze platform, which is our generative AI and agentic AI platform, and over the last three years it has matured significantly.
What we are seeing now is a clear shift in customer behavior. Enterprises are still modernising legacy applications and running their core businesses on them. That part has not gone away. But for newer initiatives, customers are increasingly open to buying rather than building. Out of a large number of initiatives, many prefer adopting a product or SaaS model first, and then deciding whether to continue with it or convert it into in-house AI engineering later.
Q2.: A decade or more ago, IT services companies were deeply embedded in running enterprise back-end systems. Has that model fundamentally changed?
Rajesh Sinha: Not completely. Core systems are still critical. Banks still rely on large engineering teams for payment platforms. Insurance companies still depend on policy administration and claims systems. Those systems are not being replaced overnight by AI.
What has changed is where innovation happens. Enterprises are not asking us to stop running or building their core platforms. Instead, they want innovation layered on top of those systems. For example, rather than building a completely new application for something like anti-money laundering, they may want an AI agent that works on top of existing systems.
This introduces more choice. Customers now have access to multiple AI agent vendors and platforms. As a result, competition has increased, but so has opportunity. In the end, the vendors who truly understand the customer’s business and data environment are the ones who can deliver meaningful innovation.
Q3: Given this shift, how has Fulcrum changed the way it approaches enterprise customers?
Rajesh Sinha: Earlier, the conversation was very services-driven. We would understand the problem, define requirements, propose a solution, and then execute through a services engagement. Today, customers want to see and experience solutions before committing.
With Ryze, when we meet a financial services or insurance client, we can showcase multiple AI agents that we have already built. These are domain-specific agents, not generic demos. Customers can upload their own data, run models, see reports, and understand what is possible before deciding whether to buy or build. This is a big change. It is more product-centric. Customers want to feel confident about the solution before they invest further, and this approach helps them visualise outcomes early.
Q4: What is the significance of the Ryze platform in this context?
Rajesh Sinha: Ryze has matured into a foundational platform for building agentic AI journeys. At a basic level, it provides the building blocks required to create enterprise-grade AI agents. There are multiple layers involved. First is the infrastructure layer, including compute and security. Second is the data layer, because AI works on vectorised data rather than traditional row-and-column databases. Third is orchestration.
Most business problems cannot be solved using a single large language model, so Ryze orchestrates multiple LLMs (Large Language Models) within the customer’s environment and firewall. The next layer is integration with existing enterprise systems. AI agents must connect with platforms like SAP, Salesforce, and internal systems that companies have built over years.
Finally, there is an integration server layer that enables secure data exchange between agents and systems. Together, these components allow enterprises to build and scale AI agents without compromising security or governance.
Q5: How do you describe the maturity journey of AI agents within enterprises today?
Rajesh Sinha: Most companies are currently at what I would call level two. Level one is conversational AI, such as basic customer service agents. While, level two is where AI agents begin to execute business processes and replace specific manual tasks in areas like finance, marketing, or operations.
Level three involves more advanced autonomous agents, where multiple agents work together as a system with minimal human intervention, though governance and human oversight remain important. Beyond that, there are higher levels involving general intelligence and super-intelligence, but most enterprises are not there yet.Right now, the real work is happening between levels two and three.
Q6: From a business perspective, what opportunities does this create for Fulcrum?
Rajesh Sinha: There are two major opportunities. First, onboarding speed. When customers see working AI agents and like what they see, becoming a vendor becomes faster compared to traditional services engagements.
Second is acceleration. Once trust is established with one or two agents, customers often want to scale quickly and build many more. For example, we are working with a large life insurance company that has thousands of human sales agents. Many of them have limited formal education and struggle to explain complex insurance products.
We are enabling them with an AI-powered sales agent that provides real-time information during customer interactions. This improves the quality of conversations, shortens sales cycles, and helps customers make better-informed decisions.
Q7: What are the key challenges enterprises face when adopting AI at scale?
Rajesh Sinha: One common challenge is expectations around accuracy. Many customers initially expect AI to deliver 100% accurate results. A lot of time goes into explaining that large language models work probabilistically and that accuracy improves through architecture, reinforcement learning, and techniques like retrieval-augmented generation.
Internally, this creates technical challenges. Our teams continuously refine models, write additional code, and evolve the platform. The way we approached RAG (Retrieval-Augmented Generation) six months ago is already different today. Another internal challenge is training. Product engineering teams have deep AI expertise, but implementation teams need to reach the same level of understanding for us to scale efficiently. Continuous training is critical.
Q8: How do regulation and data governance factor into your AI deployments?
Rajesh Sinha: We assume strict regulation from the start. Security and governance are built at the problem-definition stage, not added later. We define who can access data, where models are hosted, and whether data leaves the organisation.
When AI is deployed inside an enterprise, large models effectively become smaller, private models. This allows us to implement strong governance and be prepared for audits or regulatory scrutiny, regardless of geography.
Q9: Looking ahead to 2026, how are you shaping Fulcrum’s growth strategy?
Rajesh Sinha: In addition to traditional metrics like customers and revenue, we now track how many AI agents we deploy and how quickly adoption grows within accounts.
Internally, we are also using AI extensively. Today, around 25 to 30% of our code optimisation is driven by AI tools, and we aim to increase that significantly. Every department is adopting AI, whether through Ryze or third-party tools, to improve efficiency. On the market side, SaaS-based pricing allows us to reach customers with different budgets and maturity levels, helping us expand faster.
Q10: Finally, how do you see AI impacting hiring and human capital at Fulcrum?
Rajesh Sinha: We are bullish on hiring. Demand for AI engineering and domain expertise is increasing, not reducing. Over the next year, we plan to hire around 800 to 900 people, while also deploying another 400 to 500 AI agents across the business.
AI can optimise coding time, but it cannot replace domain understanding or business imagination. Software engineering as a discipline is not disappearing. AI accelerates parts of the delivery cycle, but humans still define the business objectives and stitch everything together.
(Watch this entire conversation with Rajesh Sinha, Founder & Chairman of Fulcrum Digital on TechHerald YouTube channel)