Karpathy: from vibe coding to agentic engineering

Why it matters

Karpathy coined the term 'vibe coding'. In this 2026 Sequoia conversation he pushes the idea one step further: from casually letting AI write code to disciplined 'agentic engineering' — how to bound an agent's scope, how to split work between human and machine, and when to trust versus when to keep a human hand on the brake. Half an hour that captures the real dividing line as AI coding moves from toy to production.

Key takeaways

• Vibe coding suits exploration and prototypes: fast natural-language iteration at the cost of quality and control.
• Production requires 'agentic engineering': clear task boundaries, verifiable intermediate artifacts, instant rollback.
• The human-machine split: AI generates and retrieves; humans judge, set taste, and make the final call.
• Agent permissions should open gradually: read-only, then suggest, then execute — trust is earned on evidence.
• The engineer's future edge: specifying precisely what you want, and recognizing whether it's right.