As artificial intelligence becomes a core element of enterprise strategy, many companies are racing to adopt the latest tools—chasing faster GPUs, larger models and more automation. But according to Krishna Subramanian, Co-Founder and President of Komprise, the true differentiator isn’t raw compute power—it’s how well you manage your data.
With her deep insight, Subramanian dives deep into the intersection of AI and data strategy, revealing why unstructured data is emerging as the crown jewel of enterprise AI. She also unpacks the growing importance of smaller, more focused language models, and paints a realistic picture of what the future of work looks like in an AI-driven world.
Key Points: Small Language Models and Data Management
SMLs: Small and mid-sized language models are the future of enterprise AI.
The importance of data management: data management—not just compute power—is critical for AI success.
The crown jewel: Unstructured data is the crown jewel for enterprise AI.
The human role: AI will be a tool, not a replacement, for human roles.
Select Quotes: “It’s going to be human-in-the-loop”
From the rise of smaller, smarter models to the underestimated role of data infrastructure, here are three key quotes from Subramanian that provide insight into where enterprise tech is headed.
The Rise of Smaller, Smarter AI Models
“There’s this whole movement towards small language models—things that can just run on your phone or can run in a self-driving car. It has fewer parameters, but actually it’s more accurate because it’s more purpose-built for that purpose.
“So it may not beat you at chess, but it’ll not make a mistake when you’re driving your car, which is all you care about. For enterprises, we’re going to see something—I’m calling it MLMs, mid-size learning models—which will run in enterprise data centers. It’ll run on corporate data. It’ll be very task-specific. It’ll be much more accurate and won’t have the security concerns of running something somewhere else.”
Why Data Management Is the Real AI Battleground
“If you are a business and you want to gain a competitive advantage using AI and you just use a pre-trained model, it’s not really going to do anything more for you than it would for anyone else, because it doesn’t have access to your data.
“So what’s really different—how you can use AI to your advantage—is by using your data with AI. And that’s why Gartner found that more than performance or storage capacity, it’s data management that’s really going to make the difference. And given that 90% of that data is now unstructured, this is why unstructured data is sort of the crown jewel for AI.”
AI Won’t Replace You—It’ll Report to You
“Right now people are thinking AI will do this and humans will do that, and I don’t think that’s what’s going to happen. It’s going to be human-in-the-loop.
“AI is going to assist you with things and then a human’s going to verify. So imagine you, James—you’re going to have 55 assistants doing different things for you, and you are still going to coordinate it, you’re going to manage it. But you’re going to now have all these extra resources at your disposal.”
