The internet ruined customer service. AI could save it.
TLDR
- AI collapses the cost of high-quality attention, enabling every scale business to deliver concierge-level customer experiences that were previously only viable at luxury ARPU.
Key Takeaways
- The scale/concierge split has governed consumer business for decades: Amazon and Delta get you logistics at low cost; Hermès gets you a personal relationship at high ARPU.
- The bottleneck was never intent but economics: hiring, training, and retaining enough humans to give every customer real attention does not pencil at low ARPU.
- AI agents (a16z portfolio company Decagon, founded by Jesse Zhang and Ashwin Sreenivas) are achieving 80%+ ticket deflection rates at enterprise customers including Avis, Hertz, Cash App, Mercado Libre, and Oura Health.
- The Jevons paradox applies: when cost of quality attention drops dramatically, latent demand for concierge service at scale businesses is large enough that total consumption rises, not stays flat.
- When an AI agent holds full customer history and runs continuously, the line between support and commerce dissolves; proactive recommendations become indistinguishable from sales.
Why It Matters
- Scale businesses like Delta and Verizon have never wanted bad customer service; they structurally could not afford good service at volume until now.
- Decagon’s early results show deflection above 80% alongside higher customer satisfaction, which is the first concrete evidence the concierge-at-scale model works outside luxury.
- If every scale business can afford concierge economics, “customer service” stops being a cost center and becomes a direct revenue lever embedded in every customer interaction.
Sarah Wang, Andreessen Horowitz · 2026-03-18 · Read the original
