Sarah Wang and Kimberly Tan host Jesse Zhang and Ashwin Sreenivas, co-founders of Decagon, to discuss the company's approach to building enterprise AI applications, focusing on open-source models and their deployment in large companies.
Key points
Decagon's AI agents are designed to handle all customer interactions, both reactive and proactive, acting as the business's front door.
Decagon initially used frontier models for their AI but shifted to open-source models to reduce latency and improve control over specific tasks. 2:00
The company has developed a research team to fine-tune open-source models, which now handle 90% of their workflow, optimizing for latency and cost. 3:01
Decagon's approach allows smaller, fine-tuned models to outperform larger, state-of-the-art models on specific tasks, providing better performance, speed, and cost efficiency. 5:20
Decagon's AI agents are not just for customer support but are designed to follow business processes, making them versatile for various enterprise tasks. 49:25
The company emphasizes a product-driven approach, ensuring that their AI solutions are scalable and not just custom-built for individual clients. 27:00
Decagon's founders spend significant time on sales, emphasizing the importance of understanding enterprise needs and rapidly iterating on product offerings. 45:07
The company has expanded internationally, driven by customer demand and the ease of adapting AI models to different languages. 61:01
Decagon believes AI will transform jobs but not eliminate careers, as automation allows employees to focus on more complex and valuable tasks. 78:50
Decagon's AI agents are designed to improve over time, using feedback from real-world interactions to enhance their capabilities. 32:21
Quotes
An AI agent should just be the front door of your business.
AI will kill jobs but not careers.
Decagon's approach is a glass box, not a black box.