AI Copilots Are a Dead End. Here's What Actually Works | Kavak CEO

a16z Deep Dives · 2026-03-05 · 71 分鐘
https://www.youtube.com/watch?v=eQ3d2jNN3uE影片總結
Kavak uses AI agents for 90–95% of customer interactions after a flat 2023, then grows four times.
Kavak CEO Carlos García Ottati explains how a fraudulent car sale exposed Latin America’s unsafe, informal used-car market, where 40% of transactions involve fraud and only 5% of Mexico’s used cars are financed. Founded in 2016, Kavak built four businesses beneath its marketplace—reconditioning, warranties, financing, and logistics—before expanding beyond Mexico City. Its biggest transformation came with AI: after employees failed to adopt copilot tools, Kavak deployed specialized agents for pricing, underwriting, repairs, and customer edge cases. The transition caused performance to deteriorate and kept the company flat throughout 2023, but agents eventually reached 1.5 times the performance of its best humans. Today, AI handles 90–95% of user interactions, helping Kavak grow four times and become profitable. Ottati’s operating philosophy is to remove one customer friction every week, reassess decisions within 24 hours, and periodically “fire” himself to determine whether he still fits the company’s next stage.
章節
- 0:00Kavak's scale: 10,000 monthly transactions, 3,500 employees, and operations across Latin America
- 4:58How his parents stayed optimistic through every move — the Venezuela village car that only went in reverse
- 8:05How his parents stayed optimistic through every move — from Venezuelan exile to empathy and win-win relationships
- 13:05How getting defrauded selling a car became the idea for Kavak — a 40% fraud market
- 17:34How getting defrauded selling a car became the idea for Kavak — building four businesses beneath the marketplace
- 22:46Why they stayed in Mexico City for four years before expanding anywhere — concentrating 60% of Mexico’s opportunity
- 28:00Why they stayed in Mexico City for four years before expanding anywhere — the 2017 robot pipeline and information highway
- 31:13Why they stayed in Mexico City for four years: pricing 60,000 SKUs with AI and learning from warranty failures
- 35:31Humans with blood vs. agents with electricity: Kavak’s agents solve car-breakdown edge cases
- 40:53The year everything got worse before it got better: Kavak went flat in 2023 during its AI transition
- 45:25The year everything got worse before it got better: from 10,000 employees and 99.9% market collapse to profitable growth
- 50:31The year everything got worse before it got better: solving real Latin American needs with patient compounding
- 55:09The year everything got worse before it got better: remove one user friction weekly and reassess every 24 hours
- 58:55Treat problems like a protein bar: pressure is earned through expectations
- 1:00:38Why Carlos fires himself every year: Rehiring himself for the company’s next phase
- 1:05:52Why Carlos fires himself every year: From mowing lawns at 12 to founding Kavak
這是 Tier 1 公開摘要
每章重點、段落總結、心智圖由分享者控制是否公開。想看完整分析?自己提交一支。
同頻道的其他分析
Mintlify and the Transition From Human Docs to Agent Infrastructurea16z Deep DivesMintlify’s Han Wang and Hahnbee Lee explain eight pivots, a two-day prototype, and docs becoming AI infrastructure.
To Regulate AI Effectively, Focus on How It’s Useda16z Deep DivesMartin Casado argues AI laws should target illegal use, while US regulatory uncertainty pushes startups toward Chinese open-source models.
How Palantir Scaled: Why the Best Software Is Built Backwardsa16z Deep DivesPalantir’s Akshay Krishnaswamy explains FDEs, backward product building, Foundry, and avoiding the consultancy trap.
Inferact: Building the Infrastructure That Runs Modern AIa16z Deep DivesInferact founders explain vLLM’s rise, 500,000-GPU scale, and a universal open-source inference layer
Temporal CEO on AI Agents & The Future of Software | Deep Dives with a16za16z Deep DivesTemporal powers OpenAI Codex and Snap at scale, bringing durable state, recovery, and RPC to long-running AI agents.
相關主題的分析
The Next Wave of AI Companies in Latin America | Tako Founder on The a16z Showa16z Deep DivesSebastian Mejia explains Rappi’s rise and Tako’s AI plan for Brazil’s $20–30B labor complexity.
股市暴跌,無人消費:AI贏了,但白領消失了?!《全球智能危機》Better Leaf 好葉《2028年全球智能危機》警告AI裁員、13兆美元房貸與SaaS死亡螺旋將重塑經濟
5個看似負責任,卻使你默默變窮的理財習慣!第4個最多人中招,你也中了嗎?Better Leaf 好葉好葉拆解5種過時財務習慣,從通膨、好債到薪資與資產,教你升級致富思維
Are Tariffs Fueling the AI Race? | Deep Dives with a16za16z Deep DivesJacob Helberg outlines America’s AI race strategy: superior models, global market share, secure supply chains, and tariffs.
停滯性通膨要來了?失業上升、物價飛漲、利率不降:普通人最危險的時刻到了Better Leaf 好葉美國非農減少9.2萬人、CPI仍升2.4%,高利率與AI裁員恐引發停滯性通膨