Unlike basic AI tools that summarise emails or schedule meetings, AI agents can autonomously carry out multi-step tasks from start to finish. Das pointed to financialanalysis as one of the first professional areas facing disruption. Agents are already compiling datasets, building financial models, and generating scenario forecasts — work that would typically consume an analyst’s entire day. “Companies will have to adapt,” Das said plainly.
Despite this momentum, Das acknowledged that the technology is still maturing. Human oversight remains essential for now, as agents can make errors that require review. He predicted that trust will build gradually as agents prove reliable across more use cases, including customer service and software development.
AI-native companies hold the edge
Das highlighted a growing divide between startups built around AI from day one and older companies scrambling to integrate it. AI-native firms are already operating leaner, with fewer engineers and outcome-based pricing models that bill clients per completed task rather than flat subscription fees. A focused CEO, he said, is the decisive factor separating those making the transition from those falling behind.
Running costs remain a significant obstacle, however. Because agents often operate continuously for hours or even days, computing expenses can escalate quickly. Das was candid about the stakes: “I don’t think everybody will make it. The companies that will make it are the ones that take a radical approach” — redesigning their entire organisations around AI agents, not just bolting the technology onto existing workflows.