Karpathy at Sequoia AI Ascent: Three New LLM Horizons and the Agent-Native Economy

Andrej Karpathy's Sequoia AI Ascent fireside chat articulated three new LLM capability horizons—menugen, install.md skills, and LLM knowledge bases—plus a theory of LLM jaggedness tied to domain economics, and the 'agent-native economy' framing where agentic engineering becomes a distinct skill set.

1 min read|agenticonsult Intelligence

Karpathy at Sequoia AI Ascent: Three New LLM Horizons and the Agent-Native Economy

Andrej Karpathy's fireside chat at Sequoia AI Ascent 2026 laid out three capability horizons beyond "speeding up what already exists": (1) menugen—apps fully replaced by LLMs, no classical code needed; (2) install.md skills replacing install.sh—LLMs as advanced English interpreters for software installation that debug inline; and (3) LLM knowledge bases enabling computation over unstructured data from arbitrary sources. Karpathy also articulated a jaggedness theory—LLM capability spikes correlate with domain verifiability and economic TAM—and framed "agentic engineering" as an emerging distinct hiring category. He said he has never felt more behind as a programmer.

Why It Matters

Karpathy's frameworks consistently shape practitioner thinking: the menugen, install.md, and knowledge-base horizons give teams a roadmap for where agentic investment has the highest expected return in 2026–2027.

This breaking-news item was assembled from the cited primary source with AI assistance. It is intended for rapid situational awareness — refer to the original publication for the definitive statement.

Karpathy at Sequoia AI Ascent: Three New LLM Horizons and the Agent-Native Economy

Karpathy at Sequoia AI Ascent: Three New LLM Horizons and the Agent-Native Economy

Why It Matters

Live Intel Feed