Jensen says we’ve hit the singularity. I get why he’s saying that. But the real question is what does that actually mean in practice. From where I sit, it is not some sci fi moment where machines… | Reuven Cohen

Reuven Cohen challenges the narrative around achieving artificial general intelligence (AGI), reframing what "hitting singularity" actually means in practice. Rather than a sci-fi moment where machines become self-aware, Cohen describes a more immediate and pragmatic shift: the limiting factor in building with AI agents has moved from "can we build it?" to "can we understand what we built and why it works?"

Over the past year, Cohen has been able to construct sophisticated systems—complex multi-agent workflows, self-learning loops, production deployments—without hitting hard technical walls. The constraint is no longer tooling or raw AI capability. Instead, the real challenge is comprehension: explaining high-dimensional structures, continuous adaptation, and emergent behaviors that he can observe and validate but struggles to explain cleanly.

This creates a critical gap between what AI systems can do and what humans can understand about their operation. Cohen notes that when people dismiss AI outputs as "slop," it's often not a capability failure—it's a cognition gap. They cannot parse what the system is doing.

At a societal level, we may not have reached true singularity yet. But for individual builders working with agentic systems, it's already here: the ability to move reliably from idea to working system without technical obstacles. This shift changes what matters most. Cohen emphasizes three things become essential: control over these systems, auditability (being able to trace why they made certain decisions), and knowing when not to trust what you've created. This points toward an emerging discipline combining software engineering principles with quality management practices—what some call "agentic engineering."