David Tafur, YaVendio | A Step-by-Step Guide to Building an AI Startup in LatAm

The audio and video are in Spanish. These notes are a translation.

Today I spoke with David Tafur, co-founder and CEO of YaVendio, an AI-powered virtual salesperson that helps e-commerce businesses sell through WhatsApp and social media. Its conversational technology achieves conversion rates of 5% to 15%, well above the 1% to 3% of traditional channels.

Youtube Timestamps*

[1:35] - Personal Story and Origins in the Startup World

David combined three passions from childhood: science, sports, and business.

How did his early experience selling sodas and competing in swimming shape his entrepreneurial approach?
What was his first foray into the startup world, and what did he learn from that initial failure?

[2:38] - The Birth of YaVendió and First Steps

David co-founded an artificial intelligence product agency as a side project.

What findings led him to focus on solving conversational selling for e-commerce?
Where was the company at that point? How much capital would you say is needed to start an AI startup?

YaVendió’s first steps consisted of multiple discovery conversations.

How did he approach those interviews with companies across different sectors?

[16:22] - Minimum Viable Product and User Experience

YaVendió’s initial MVP was a chat that interacted with the consumer, with almost no interface for the business.

How did David validate this first prototype with real customers?
What lessons did he take away about building fast and not obsessing over early perfection?

As they moved forward, they discovered that the real differentiator was creating a highly specialized “virtual salesperson.”

Why not think of it only as software, or only as a virtual employee, but as a third path?
How do they justify their high monthly price by showing clear returns (headcount reduction, sales growth, etc.)?

[35:39] - Building an AI Startup in Latin America

For David, starting an AI company does not require a large amount of initial capital, but focused talent.

How did his prior experience in AI, and that of his co-founders, speed up the building of YaVendió?
What advice does he give to people starting an AI startup about the balance between raising capital and moving forward with their own resources?

The costs and the evolution of AI models change so fast that it is key to anticipate them.

Why is it worth not ruling out features that seem expensive today and could be much cheaper in a few months?
How does this influence the roadmap of an AI product?

[36:41] - Product Strategy and Technical Aspects of AI

YaVendió combines several models and evaluations to achieve reliable conversational sales.

Why is a ChatGPT wrapper not enough to solve the complexity of selling in e-commerce?
How does David handle the “third layer” of AI evaluations and extensive testing to ensure quality in the responses?

A scientific mindset is essential:

How does the team drive experimentation and the constant adoption of new tools?
Why are testing pipelines and development automation essential in AI?

[45:21] - Competition and Market Niches

The global market for AI sales agents is full of competitors, but in Latin America there are still gaps.

Why does David believe the way people sell in the region is very different and offers differentiated opportunities?

YaVendió’s focus is on deepening the automation of several “virtual salespeople” for e-commerce.

[53:28] - Vision for the Future and Final Reflections

In the long term, David imagines a “business in a box” enabled by AI and run by very few people.

How could solo founders or microbusinesses use multiple AI agents to scale enormous sales volumes?
What is the potential impact on the structure of retail commerce in the region?

David says he is fascinated by the chaotic speed of AI’s evolution.

What does he think of the rapid drop in costs and the constant emergence of new models?
How does he believe this speed will keep reshaping the innovation landscape in Latin America?