Your AI Is Only as Smart as Your Data Fabric

By the end of 2025, half of all companies had AI running in at least three business functions—finance, supply chains, HR, customer ops. Copilots, agents, predictive systems, you name it. We’ve moved past the experiment phase. AI is embedded.

But here’s the thing nobody wants to admit: the models aren’t the bottleneck anymore. The data is. Not the volume, not the speed—the context. AI can crunch numbers and spit out answers faster than any human, but if it doesn’t understand why that data matters, it’ll make confident, fast, wrong decisions.

Irfan Khan, who runs SAP’s data and analytics product side, put it bluntly: “AI moves fast, but without context it can’t exercise good judgment. Speed without judgment doesn’t help. It can actually hurt us.”

He’s right. And he’s talking about something most enterprise architects have been ignoring: the data fabric isn’t just a plumbing problem anymore. It’s a judgment problem.

The Context Tax

Traditional data strategies were built for reporting. You extract from operational systems, dump into a warehouse or lake, build dashboards. That worked fine when humans were in the loop—they brought the missing context. They knew which customer was strategic, which contract clause mattered, which tradeoff was acceptable during a shortage.

But now we’re asking AI to act on that data autonomously. And the data, stripped of its original business meaning, is just noise.

Khan gave a good example: two companies both using AI to manage supply-chain disruptions. One feeds in raw inventory levels, lead times, supplier scores. The other adds context—strategic accounts, acceptable tradeoffs, extended supply chain status. Both systems run fast. Only one makes the right call.

That’s the “context premium,” as he calls it. And it’s real money.

The Numbers Are Ugly

Most companies know they’re not ready. Only one in five considers their data approach “highly mature.” Only 9% feel fully prepared to integrate and interoperate their data systems. That’s a staggering admission from people who are already deploying AI in production.

You’re flying the plane while you’re still building it, and the engine is held together with duct tape.

Don’t Consolidate, Integrate

The fix isn’t another data lake. It’s a data fabric—an abstraction layer that sits across all your infrastructure, clouds, and operational systems, preserving the semantics. For agentic AI, this fabric becomes the primary interface. Agents query business knowledge, not raw storage.

Knowledge graphs are a big part of this. Instead of asking “what’s the inventory level?” an agent can ask “which strategic accounts are at risk given current inventory constraints?” That’s a fundamentally different capability.

It’s not about moving everything into one place. It’s about connecting everything while keeping the meaning intact. That’s harder, but it’s the only path that doesn’t end with AI making expensive mistakes at machine speed.

What This Actually Means

If you’re building AI systems today, stop obsessing over model architecture and start obsessing over your data semantics. Your LLM or agent is going to act on whatever you feed it. If you feed it context-free data, you get context-free decisions.

And in a world where AI is making autonomous decisions across finance, supply chains, and customer operations, context-free decisions aren’t just unhelpful—they’re dangerous.

The data fabric conversation has been around for years, mostly as infrastructure talk. It’s time to reframe it as an AI readiness conversation. Because the models are ready. The data isn’t.