Summary
The episode examines how Kavak rebuilt its company around AI agents, moving from under-adopted copilot tools to autonomous agents that now handle roughly 90–95% of customer interactions. Carlos García Ottati explains why operating in Latin America required vertically integrating multiple businesses—e-commerce, reconditioning/warranty, financing, and logistics—beneath a single consumer-facing product to solve high fraud, scarce financing, and weak payment rails. The conversation covers the operational challenges of deploying AI at scale: building ontologies, data pipelines, and safety 'brakes,' and accepting a year of flat growth while restructuring. It also digs into founder-level lessons about re-entering operational roles during transitions and intentionally adopting new leadership personas to meet the company’s evolving needs.
Key Takeaways
- 1An agent-first approach can supplant copilot tools and scale customer interactions.
- 2In emerging markets, vertical integration is often necessary rather than an asset-light platform model.
- 3Operationalizing AI requires foundational engineering: ontologies, data pipelines, and deployment controls.
- 4Major transformations require patience and acceptance of short-term pain for long-term gain.
- 5Founders may need to 'hire themselves back' and adopt new leadership personas during critical transitions.
Notable Quotes
"40% of used car transactions in Latin America end fraud."
"In Mexico, only 5% of the used car market is financed, compared to 90% in the US."
"We replaced co-pilot tools that employees wouldn't adopt with agents that now handle more than 90% of customer interactions."
"We had to build four different businesses below our consumer experience to make this work."
"The pros of hiring myself back to do that is one, there's a huge probability that I'm going to care more than anybody else about it."
"So like you can probably like see certain things and solve them quicker than anybody that you're going to bring in."
"And in order to build businesses, you have to become a persona."
"An instant is really important in decision making. But now it's guided towards a framework of what the company needs from you in this stage..."
Episode questions
Why did Kavak move from copilot tools to autonomous agents?
Employees weren’t adopting copilot tools and the company needed consistent, scalable handling of complex, high-variance customer cases. By building ontology and data infrastructure first, Kavak could deploy agents funnel by funnel and ultimately achieve parity and then superiority over humans for many tasks.
What structural market problems made Kavak build multiple businesses under one consumer product?
Emerging-market gaps—high fraud rates, near-absent financing (5% in Mexico), limited payment rails, and weak logistics—meant Kavak had to build e-commerce, reconditioning/warranty factories, financing, and logistics to make used-car commerce safe and scalable.
What was the operational cost of transitioning to AI agents?
Kavak experienced about a year of flat growth while restructuring; short-term KPIs and customer experience dipped as systems replaced human workflows, but the company persisted to achieve improved scale and profitability thereafter.
How did Kavak choose which markets/cities to expand into?
They prioritized large, dense cities with deep problems to solve (fraud, low financing penetration) and ticket economics (>~$10k) to ensure unit economics; expansion followed a pattern of building infrastructure and data capability in-market for ~4 years before scaling.