AI can thrive on limited data, India doesn’t need to compete on volume: Sachin Panicker, Fulcrum

Artificial Intelligence (AI) is evolving fast, but many businesses are still figuring out how to use it effectively. While AI tools are becoming more powerful, challenges like integration, data availability, and workforce adaptation remain. 

In conversation with TechCircle, Sachin Panicker, Chief AI Officer at Fulcrum Digital, breaks silence on these challenges and opportunities. He explains how agentic AI builds on traditional AI, why businesses struggle with adoption, and how India is tackling data scarcity. He also discusses the future of work, whether AI will replace jobs or create new ones, and the growing demand for roles like AI consultants and prompt engineers. Edited Excerpts:

Can you briefly explain how agentic AI differs from traditional AI? 

Agentic AI builds upon traditional AI rather than replacing it. Think of it as an additional layer on top, while it introduces new capabilities, it still relies on conventional AI at its core. 

At its foundation, agentic AI uses a large language model. It’s not a completely separate system but rather an evolution of traditional AI, with significant overlap between the two. The key difference lies in how it utilises traditional AI components. 

Agentic AI stands out because it can anticipate, plan, reason, and take action using tools, both in the short term and the long term. When it arrives at a final decision or action, it leverages tools to gather information and then employs a large language model to generate and deliver a response. 

So, while agentic AI is rooted in traditional AI, its architecture is more advanced, enabling it to do far more than conventional AI.  

Where do enterprises fit in? Since many use your AI agents, what specific challenges do they face in adoption? 

There are countless platforms available today, and nearly everyone is on a journey with agentic AI. It’s worth mentioning that back on October 11, 2023, during an AI summit in New York, I introduced the term “AI agents.” At that time, no one was really talking about them. Since then, we’ve been building on that foundation. 

For enterprises, the key challenge is how to integrate AI agents into their platforms in a way that simplifies adoption. How can businesses bring in an agent and put it to work seamlessly? That’s where our platform design stands out. Unlike traditional systems that require deploying an entire platform and undergoing extensive customisation and integration, our approach is streamlined and easy to use. 

Take the insurance industry as an example. Suppose a company wants to speed up its claims management process, which currently involves ten steps. With our platform, they can reduce that to nine by having an AI agent handle one of the steps. All the company has to do is deploy that specific agent for the task. 

What sets our technology apart is its ease of implementation. Instead of spending weeks integrating AI into their workflow, businesses can simply purchase and deploy it. While minor adjustments may be needed to align with their infrastructure, they can start using it almost immediately sometimes as soon as the next day. 

Traditional AI agents often present integration challenges, making it difficult to fit them into existing business processes. However, we’ve solved this problem by decoupling agents. Each agent functions independently, allowing businesses to select and integrate only the ones they need.  

With AI’s rise, there’s talk of data scarcity in India vs. the West. Given your work with AI agents, where does India stand? Is data scarce, structured, or unstructured, and how are you using it?

You're absolutely right, data is scarce. Even when it comes to companies like OpenAI, there’s still a lot of speculation about how they acquire their training data. I won’t get into that, but speaking for Fulcrum and FD Ryze, we’ve managed to build our models despite the limited availability of data. 

In reality, you don’t need vast amounts of data. I’ve always advocated that working with smaller but high-quality, clean data is far more effective. A large volume of junk data only creates more problems. Today’s AI models can be trained with relatively small datasets, as long as the data is well-curated and relevant. That’s exactly what we do, we rely only on publicly available data, ensuring strict compliance with privacy and security standards. 

Security is a cornerstone of our approach. While we have deep enterprise and domain knowledge, we do not acquire data from customers. Their data remains private. Instead, with our FD Ryze LLM, we generate synthetic data tailored for specific applications. 

For example, if you’re working on fraud detection in the banking sector, you need a substantial dataset but real-world data may be scarce. In such cases, we start with whatever publicly available data exists and then use our FD Ryze LLM to generate high-quality synthetic data. This allows us to train AI models effectively, and the results have been outstanding. 

So, don’t be discouraged by data scarcity. Use what’s available, and where needed, create your own high-quality synthetic dataset. While the process is complex, it’s entirely possible and it works. 

With AI automating tasks, job losses are a concern. Given India's skills gap, will AI replace jobs or complement human intelligence? 

Many Western CEOs have expressed concerns, though their perspectives are often influenced by vested interests in AI-driven products. But I don’t believe AI will replace human jobs entirely. In India, we don’t need to fall into fearmongering. Instead, we should see AI as a partner, a tool for innovation and productivity. Embracing AI will allow us to focus on what we, as humans, are truly good at creative and strategic thinking. 

Think about it, every day, we spend so much time on mundane tasks—laundry, cooking, household chores that drain our energy. If AI and automation can handle these, we’ll have more time for meaningful work that can drive progress and make the world a better place. 

Now, will AI take jobs? Not exactly. Rather than replacing jobs, AI will augment human capabilities. Take customer service, for example. Today, a company might need 100 call center agents. As AI becomes more efficient, businesses may not need to hire 200 more agents in the future. Instead, AI will assist the existing workforce, making them more productive. This shift is inevitable just like how past technological advancements, from the printing press to the telephone, transformed industries without eliminating human work entirely. 

AI has immense potential in areas like drug discovery, weather forecasting, and countless other fields. The key is to integrate AI into our workforce while ensuring people reskill themselves to work alongside it. The future belongs to those who understand and utilise AI effectively. 

That’s why education needs to evolve. Colleges and universities must update their curricula to align with technological advancements so that graduates are equipped to use AI in their professions. While AI may reduce demand for certain entry-level jobs, it will also create new opportunities. 

For example, roles like prompt engineering are emerging. Some say this field will disappear soon, but I disagree. Writing effective AI prompts is a niche skill that requires deep understanding. As an AI scientist, I can craft a far more effective prompt than someone without expertise. Training people in these skills presents a massive opportunity. 

Ultimately, AI will reshape the job market, but it won’t eliminate the need for human workers. Instead, it will create new roles and industries that we haven’t even imagined yet. There’s no need to worry we just need to adapt. 

What emerging roles do you see evolving in the future? 

Prompt engineering is a great example of how AI is evolving. Another key area is AI adoption in enterprises. Right now, businesses have barely scratched the surface of AI’s true potential. Many people don’t fully understand how to leverage AI effectively. 

This gap presents a major opportunity. AI consulting—helping businesses understand and implement AI—will be crucial. Have you spoken about this in relation to MSME sectors? Most business owners only see the tip of the iceberg when it comes to AI. They are unaware of its full power and how to harness it for growth and efficiency. 

This is where young graduates come in. Engineering graduates, for example, have the analytical mindset and problem-solving skills needed to bridge this gap. But this isn't just for engineers—graduates across various disciplines can step into these roles. They don’t need to develop AI from scratch; instead, they need to learn how to use AI to enhance productivity and efficiency. 

By equipping themselves with AI knowledge, they can become first-level consultants for MSMEs and other businesses. There’s an entire layer of jobs waiting to be filled millions of opportunities in AI consulting and implementation. And yet, very few are currently doing this. The future belongs to those who seize these opportunities. 

---