Descriptions:
The Cognitive Revolution hosts Nathan and Pash open with a deep dive into leaked intelligence around Anthropic’s internal “Model 2” — referred to in some circles as Mythos 2 — sourced from Dylan Patel of SemiAnalysis. Drawing on Anthropic’s own redacted risk report, they walk through the model’s performance on the internal CoBench metric (roughly 8.8 percentage points above Mythos Preview) and the Epoch Capabilities Index, contextualizing what those scores mean relative to Claude Opus 4 and Anthropic’s stated 85% staff-replacement threshold.
The episode then shifts to enterprise AI strategy, featuring the CEOs of Arthur AI and DataCamp. Topics include how companies are currently identifying and prioritizing high-value agentic use cases, the emerging role of agent observability and monitoring, and how engineering team structures are changing as product managers, customer success staff, and engineers alike begin generating code through AI coding agents.
The conversation closes on the long-term composition of technical teams, arguing that the most valuable future employees will be “builders” who combine product sensibility with enough technical fluency to direct AI agents effectively — regardless of whether they come from engineering or product management backgrounds. A strong episode for anyone tracking both frontier model capabilities and the practical transformation of enterprise software teams.
📺 Source: Cognitive Revolution “How AI Changes Everything” · Published August 18, 2026
🏷️ Format: Podcast
