The biggest barrier is getting the data to act on. Without context from data that’s properly masked and curated, an agent is just another tool—it won’t know what to do.
AI agents can’t fix what they can’t find. The dream of seamless automation and hyper-efficient workflows is quickly becoming a corporate obsession, but chaotic enterprise data kills it at the gate.
Sumit Chakraborty, Director of Engineering and Senior Client Partner at Brillio and former Oracle engineer, is deep in the trenches of enterprise data and AI transformation.
The data bottleneck: While agent tech is advancing fast, with growing context awareness and multi-agent capabilities, Chakraborty says the real hurdle is far more basic. "The biggest barrier is getting the data to act on. Without clean, curated, and properly masked data, an agent is just another tool—it won’t know what to do," he explains. In his view, disorganized enterprise data remains the chokepoint that could stall even the most ambitious AI plans.
Check the pipes: The vision of unified agent orchestration is gaining ground, even with messy data. Chakraborty calls the underlying tech "a frontier technology," fueled by open source momentum which could soon unlock frameworks to handle real enterprise tasks like onboarding. "As long as that plumbing is getting done, I don’t see any reason why agents can’t get in and do a lot of work by the end of the year."
The real power, the 'innovation arbitrage,' comes when you use AI to fundamentally reimagine workflows, perhaps even eliminating steps you thought were essential. That's where you unlock multiples of efficiency and true transformation.
The new MVP: Once the data plumbing is in place, Chakraborty says it’s time to aim higher than automation. His message: don’t just whitewash old processes—rethink what the minimum viable product looks like in an AI-native world. He points to code generation as a case in point: early tools automated tasks, but newer vertical solutions reshape the developer experience itself. "If, in a given domain, the top three use cases can be 65% handled by AI working alongside a human—with the right guardrails—that’s a conversation every leader should be paying attention to," Chakraborty says.
No Band-Aid fix: "Many businesses approach AI looking for 'automation arbitrage'—simply layering it onto existing workflows and personas. You'll only get so much from that," Chakraborty says. "The real power, the 'innovation arbitrage,' comes when you use AI to fundamentally reimagine those workflows, perhaps even eliminating steps you thought were essential. That's where you unlock multiples of efficiency and true transformation."
Lean vs. bloated: As enterprises lean into AI-driven innovation, the broader market is shifting too. Will we see another "cloud war" dominated by a few giants? Chakraborty doesn’t think so. "You will see a clear demarcation of the winners, and they won't necessarily be the enterprise app companies," he predicts. He sees a leaner future, where AI takes over transactions and bloated, page-heavy systems fade out. "You might just need a pervasive data store for your business, and AI will be able to operate directly within it," he says. As app and data companies blur through acquisitions, customers will face more options—shaped largely by how deep their investments already run in a given platform.
