NEARCON: A public rift over trusting agents with real capital

At a NEARCON session in San Francisco, two high-profile crypto figures staged a direct disagreement about when autonomous systems should be allowed to handle real money, spotlighting a split inside the industry over speed versus safety. Arjun Sethi presented an aggressive timetable, describing agent-like features Kraken is building and signaling belief that fully autonomous portfolio management could be plausible within six to twelve months. In contrast, Haseeb Qureshi framed the problem as a reliability threshold: he argued public demos mask brittleness and said only a tiny share of his personal capital would be appropriate to hand off now — roughly 5%. Both speakers agreed autonomous finance will arrive eventually, but they diverged sharply on acceptable error rates and operational readiness. The exchange executive emphasized rapid iteration and defensive controls scaling with new attack vectors; the investor emphasized that intermittent failures producing large tail losses remain the core hazard. The conversation included explicit signals about product timing: Kraken’s roadmap moves from prototypes to customer-facing agent features over weeks-to-months, while venture investors and security teams will keep demanding demonstrable fault tolerance at scale. That timetable creates immediate consequences for startups: those building agent primitives, observability layers, and secure oracles will see accelerated demand, while projects that underinvest in safety may face rapid de-risking. The public debate also changes signaling to customers and regulators: an exchange claiming near-term readiness invites fresh scrutiny from supervisors and amplifies consumer protection questions. Taken together, the session crystallized a practical industry fault line — aggressive commercialization versus conservative validation — that will shape hiring, funding, and product priorities over the next 6–12 months.