Co-founder of Anti Fund States Most So-called Agentic Products Are Just Interns in Trench Coats Clicking Buttons on Worse Dashboards
Geoffrey Woo, co-founder of Anti Fund, stated that most so-called "Agentic" products are merely "interns in trench coats" clicking buttons on worse dashboards.
This metaphor directly points to the current lack of true autonomous closed loops in many AI agent tools, which still heavily rely on human intervention or underlying systems, with actual effectiveness far from achieving independent agent levels.
This viewpoint drives AI development capital to concentrate on Agent frameworks that possess real autonomous cycles and production-level substitution capabilities, benefiting event-driven high-execution teams from product differentiation, while concept-hyped pseudo-Agent projects face pressure from declining user trust.
Source: Public Information
ABAB AI Insight
Geoffrey Woo has previously criticized the hype and superficial demonstrations in the AI industry. This "interns in trench coats" metaphor continues his pragmatic investment style, emphasizing that the core of Agentic products lies in quantifiable workflow substitution rather than superficial autonomy, similar to his earlier sharp observations on the herd mentality of VCs and categorization of founders.
In terms of capital pathways, early investors like Anti Fund will continue to allocate resources to Agent projects that can genuinely eliminate inefficient processes, filtering out "worse dashboard" products through rigorous validation. The strategic motive is to capture asymmetric returns and avoid capital waste on pseudo-innovations.
This reflects the industry's reflection on the evolution from early chatbots to mature agents, consistent with the current transition of AI Agents from demonstration phases to production-level closed loops.
Essentially, this is about capital concentration: real Agentic capabilities accelerate the filtering of pseudo-products, mechanism-wise concentrating development resources and venture capital from concept hype to a few platforms with deep integration and autonomous decision-making capabilities, further strengthening top Agent companies' pricing power and market dominance in enterprise workflow substitution.
ABAB News · Cognitive Law
The title of Agentic is easy to obtain, but true closed loops are hard to find. Top capital invests in those that can eliminate buttons rather than just click them. Most disguise as interns in trench coats, while few lock in autonomous cycles, leveraging actual departmental-level substitution. Selling demo screenshots may yield temporary financing, but selling true efficiency wins long-term adoption; winners always treat Agents as intern terminators rather than upgraded versions.