Inside the expert network training every frontier AI model | Garrett Lord (Handshake CEO)

Overview

This episode explores how Handshake, originally a career network for students, leveraged its vast network of academic experts to create a groundbreaking data-labeling business for AI models. In just eight months, this new venture hit $50 million in revenue and is on track to surpass $100 million in its first year. The conversation dives into the mechanics of data labeling, the shift from generalist to expert-driven AI training, and the strategies for building a startup within an established company.

Notable Quotes

- There will never be a time like this. I've never seen anything like it. I doubt I'll ever feel anything like this in business again, where there's unlimited demand. – Garrett Lord, on the unique opportunity in AI data labeling.

- The only moat in human data is access to an audience. – Garrett Lord, on Handshake's competitive advantage.

- Being AI native is like having your Iron Man suit on—young people are at a huge advantage. – Garrett Lord, on how AI tools empower the next generation of workers.

🧠 The Role of Experts in AI Training

- AI models have shifted from relying on generalist data to requiring domain-specific expertise.

- Handshake's network includes 500,000 PhDs and 3 million master's students, enabling them to provide high-quality, specialized data for AI labs.

- Experts identify weaknesses in AI models, such as flawed reasoning or incorrect outputs, and provide corrections to improve model performance.

- Examples include PhDs in physics breaking down complex problems or music students refining AI's understanding of audio.

🚀 Building a Startup Inside a Startup

- Handshake created a separate team, office, and operational structure for its AI data-labeling business to maintain focus and agility.

- The team grew from 5 to 75+ employees in just a few months, with a culture emphasizing speed, ownership, and high performance.

- Clear metrics, rigorous operating cadences, and a leave nothing to chance mindset were critical to success.

- The new business leveraged Handshake’s existing trust and relationships with universities and students, eliminating customer acquisition costs.

📊 The Shift from Generalists to Experts in AI Data Labeling

- Earlier AI training relied on low-cost, generalist labor for tasks like drawing bounding boxes.

- As AI models advanced, the need for expert-level data emerged, focusing on high-value domains like STEM, law, and medicine.

- Handshake’s decade-long relationships with top academic institutions positioned it to dominate this new market.

- The company’s ability to scale quickly while maintaining quality has been a key differentiator.

🌍 The Future of Work and AI’s Role

- AI tools are not eliminating jobs but transforming them, enabling workers to be significantly more productive.

- Young, AI-native professionals are uniquely positioned to thrive, leveraging tools to perform tasks that previously required multiple roles.

- Handshake’s data-labeling business also provides students and researchers with lucrative opportunities, such as earning $150/hour while advancing their fields.

- The broader vision includes revolutionizing job matching and hiring processes with AI, making them faster, more accurate, and less manual.

💡 Lessons for Entrepreneurs

- Handshake’s pivot highlights the importance of recognizing and leveraging existing assets in new ways.

- The company’s success underscores the value of focus, speed, and building trust with both customers and contributors.

- AI’s rapid evolution presents countless opportunities for startups to create meaningful, impactful businesses.