How Kavak Rebuilt Itself Around AI Agents | Alejandro Maza Ayala

How Kavak Rebuilt Itself Around AI Agents | Alejandro Maza Ayala – The a16z Show

Market views

Bullish

AIMODELS— AI models are rapidly improving and enabling superhuman performance at scale. The speaker demonstrates that AI agents can now outperform human experts across multiple dimensions, with conversion rates 2.1x better than humans and continuous improvement through feedback loops.
AILABS— AI labs are producing models with exponentially increasing capabilities that enable recursive self-improvement at the organizational level. The speaker destroyed a working system to rebuild around newer models, indicating confidence in continued rapid advancement.
MLINFRA— AI infrastructure enabling agent deployment at massive scale (100k-200k agents daily) represents a structural shift. The speaker emphasizes spending equal resources on evaluation infrastructure as on agents themselves, indicating this is a critical bottleneck being solved.

Bearish

TRADFI— Traditional financial services are being disrupted by AI-native companies that can approve loans in under 3 minutes versus 2+ months for traditional banks, with better personalization and customer experience. Incumbents face structural disadvantages in adopting AI deeply.
RETAIL— Traditional retail and service companies face creative destruction from AI-native competitors. The speaker argues that superficial AI adoption yields only 6% improvements while complete redesign yields 3x+ gains, creating an insurmountable competitive gap for incumbents unwilling to rebuild.

Not investment advice.