Written by: Techub News

In this in-depth conversation with a16z, Carlos García Otati, co-founder and CEO of Kavak, the largest used car trading platform in Latin America, shared his extraordinary growth journey, thoughts on building a billion-dollar business in complex markets, and why he firmly believes that the currently popular "AI co-pilot" model is a dead end. As a complex entity that started in Mexico, with operations across Latin America and the Middle East, and an annual transaction volume exceeding 100,000 vehicles—described as a mix of "Amazon + Toyota + Citibank"—Kavak positioned itself for AI several years in advance. Its transformation journey holds significant reference value for any business facing challenges in scaling, localization, and efficiency.

Turbulent Childhood and Entrepreneurial Genes: Optimism, Adaptability, and People Skills

Carlos García's growth trajectory is quite unique. Due to his father being a naval officer, he attended 14 different schools before the age of 17, with experiences across Venezuela and the United States. This extreme instability shaped his resilience in facing the ups and downs of entrepreneurship in the future.

He summarizes this experience as having provided him with three key qualities:

• Familiarity with Change: He deeply understands that “you might be at the pinnacle today, but you could fall to the bottom tomorrow, and vice versa.” This awareness helps him maintain emotional stability through the highs and lows of entrepreneurship, understanding that everything changes again, thus avoiding losing oneself in times of prosperity or collapsing in times of adversity.
• Family as a Totem of Optimism: Amid constant change, family was the only unchanging element. He particularly mentions the remarkable optimism and creativity exhibited by his parents during extreme difficulties (for example, the family being stationed in remote areas and having to only drive the one shared car in reverse). This attitude of "doing the right thing and staying optimistic no matter what" became the cornerstone of his values.
• The Ability to Assess People and Build Win-Win Formulas: Through constant changes in his environment, he learned to quickly read others and developed a formula for creating win-win relationships of "1+1=4". This enabled him to rapidly create favorable situations for himself and those around him in any new environment, a crucial ability in team building and managing partnerships.

Finding Kavak: From Personal Fraud to Solving Systemic Problems

Carlos's entrepreneurial inspiration directly stemmed from two painful personal experiences. The first was being defrauded while trying to sell a car in Colombia, resulting in a loss of the vehicle; the second was when he bought a car in Mexico through classified ads, only to be stopped by police two months later due to forged documents, almost leading to imprisonment.

This prompted him to conduct in-depth research and discover several shocking facts about the used car market in Latin America:

• High Informality: 90% of transactions occur in informal markets (person to person).
• Astonishing Fraud Rate: About 40% of transactions involve some form of fraud, going beyond just the "lemon" problem (used cars with hidden defects) to include serious crimes such as kidnapping, murder, forgery, and vehicle theft.
• Extremely Low Financial Penetration: In Mexico, only about 5% of used car transactions are financed, while in the United States, this figure is close to 90%. This means that the vast majority of people cannot afford vehicles through credit.

Carlos realized that this was not just a business opportunity but a massive social issue. In Latin America, not having a car means relying on unsafe and inefficient public transportation, leading to extremely long commutes and low quality of life. Owning a car is often a key step for families to enter the middle class and change their destiny. Thus, he established his goal: “How can we enable more people to enter the middle class through cars?”

Building a Complex Ecosystem: Why the Model Must Be "Heavy"

Based on his previous experience at the Latin American e-commerce platform Linio, Carlos understood that replicating the light platform model from the U.S. in regions with weak infrastructure would not work. To address the core pain points of used car trading (safety, trust, finance), it was necessary to build the underlying infrastructure from scratch. Therefore, from the very beginning, Kavak was a highly vertically integrated complex entity, which he likens to a combination of “Spotify (experience) + Amazon (e-commerce) + Toyota (refurbishment and warranty) + Citibank (finance).”

Specifically, Kavak built four major pillar businesses simultaneously:

1. E-commerce Platform: Online buying and selling of vehicles.
2. Factory Refurbishment and Warranty System: Establishing large refurbishment centers to standardize the refurbishment of acquired vehicles and provide after-sales warranty services. By having warranties, the company gains crucial data on vehicle failures for continuous improvement.
3. Financial Engine: In-house risk control and financing capabilities to provide consumers with car loans.
4. Logistics Network: Managing vehicle inspection, transportation, and delivery.

This "heavy" model was not favored in an era when capital sought light assets, but Carlos believed that in Latin America, it was the only way to provide reliable solutions and establish a long-term moat.

AI Strategic Revolution: From "Co-Pilot" to "Agent Operating System"

The essence of Kavak's business is handling vast amounts of "edge cases". Every car, every customer, every failure is unique. As the company grew from a few hundred to tens of thousands of employees, Carlos foresaw as early as 2017 that traditional human communication and decision-making processes would not support future growth. Shortly after the publication of the Transformer paper, he released an internal memo announcing the shift to supplement the "human pipeline" with a "robot pipeline".

His core insight is: The core bottleneck of a scaling enterprise is the flow of information. As the company grows, different departments (such as finance, parts, pricing) can no longer communicate as efficiently as during the startup phase, resulting in slower speed and lower quality in resolving customers' complex issues.

Therefore, Kavak's AI strategy was never about creating a "co-pilot" to increase individual human efficiency; instead, it was to build an "operating system" where different AI "agents" could communicate and collaborate. The goal of this system is to connect all business lines within the company based on an "ontology", allowing information to flow freely, akin to swarm intelligence.

The specific implementation path is as follows:

• First Build the "Brake System": Construct a central system that can understand all user interactions, failures, and successes, clearly defining the boundaries and methods for AI deployment.
• Start with Human-Machine Collaboration: Introduce AI agents to collaborate with human employees in specific business processes (such as a particular customer service funnel), with a "coordinator" distributing tasks. Depending on the complexity of the tasks, either a human or AI will handle them.
• Suffer Short-term Pain, Commit to Long-term Investment: When an AI agent first takes over a function, its performance typically falls below that of humans, leading to short-term declines in customer experience and sales metrics. Kavak experienced a growth plateau for nearly a year, but they had no Plan B and insisted on having the engineering team optimize it until AI matched and surpassed human levels.
• "Burn the Boats": After validating the model in a single funnel, they rolled it out across the entire company. Currently, 95% of Kavak's customer-facing interactions are handled by AI agents. Backend systems such as pricing and invoicing are also powered by machine learning, and they are building agents to manage these systems, achieving the integration of front and backend AI.

Carlos emphasizes that their technology construction is future-oriented ("We are not building for GPT-4, but for GPT-7"), proactively laying out architecture to quickly leverage new models as they emerge.

Through Crisis: The Double Pressure of Market Collapse and AI Transformation

From 2021 to 2022, Kavak's workforce surged to ten thousand employees, leading to enormous daily cash burn. Subsequently, the used car market suddenly collapsed, with transaction volumes plummeting by 99.9%, while the capital environment sharply deteriorated. The company faced a dilemma: needing to shrink for profitability, yet requiring growth to maintain economies of scale.

It was under this extreme pressure that a full pivot towards AI became an "inevitable" choice. Carlos likens this transformation to Netflix's shift from DVD rentals to streaming, comparing traditional car dealerships to "Blockbuster," where Kavak's initial e-commerce model was "DVD rentals," and the AI transition is "streaming." He told the team that not shifting towards "streaming" would lead to doom.

What supported them through this tough time was a strong sense of mission: 40% of Kavak's customers are purchasing a car for the first time in their lives. They understood that they were genuinely changing their users' and their families' life trajectories, and this sense of being "needed" provided deep motivation.

Advice for Entrepreneurs: Focus on Problems, Enjoy the Process

For entrepreneurs in emerging markets (especially in Latin America), Carlos offers the following advice:

• Address Real Problems, Not Create Demand: A large segment of the Latin American population still struggles with basic needs; entrepreneurs should focus on products that people "need" rather than just "want," creating real social impact.
• Dare to Think Big: The Latin American environment often does not encourage grand ambitions, but entrepreneurs must break out of this mindset and think about global issues.
• Eliminate 1% of Friction Daily: This is Carlos's core principle for managing the ups and downs. Focus weekly on removing a small point of friction for users; the long-term compounding effect is astonishing. He analogizes that on even the worst days, he could go wash a car himself, as delivering a clean car is the core value.
• Enjoy the Process of Problem Solving: Entrepreneurship is about constantly solving complex problems. One should view problems as "protein bars"—the source of nutrition for building muscle (competitive barriers). Stress is earned and signifies that others have expectations of you.
• Regularly "Fire Yourself": Every year or few years, he conducts an "self-firing" exercise. He imagines hiring a new CEO to realize the next phase of the company’s vision, clarifying the responsibilities, risks, and protections of the position. After weeks of reflection, he asks himself: Can I rehire myself for this job? This process forces him to detach emotionally and assess what old habits he needs to let go of and what new skills to develop to ensure he remains the best candidate to lead the company.

The story of Carlos García is not just a business case of building a giant in a complex market, but also a leadership model of how to use optimism as a shield, resilience as a sword, and integrate technology deeply into the enterprise's DNA with foresight to achieve transformation in times of crisis. His reflections on AI applications have transcended the tool level, entering the deep waters of organizational restructuring, providing valuable references for all companies seeking digital transformation.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。