Jesse Zhang - Building Decagon - [Invest Like the Best, EP.443]

Founders can accurately gauge an investor's future helpfulness by their actions during the pre-investment courtship phase. If an investor is unwilling to provide value when they are most motivated to win the deal, they are unlikely to be a helpful partner later on.

The end state for enterprise AI is a unified, conversational agent serving as the primary interface for a brand. This "digital concierge" will handle sales, support, and other interactions, potentially replacing websites and mobile apps as the main customer touchpoint.

While objective studies on AI coding assistants are mixed, their enterprise ROI is easily justified. Executives approve the investment because their most valuable employees—engineers—report significant productivity gains, making the business case simple regardless of hard data.

The objective, high-pressure environment of competitive math contests cultivates a first-principles problem-solving mindset and intense competitive drive. This background proves to be a powerful, if unconventional, training ground for founders navigating the hyper-competitive AI startup landscape.

The main obstacle to deploying enterprise AI isn't just technical; it's achieving organizational alignment on a quantifiable definition of success. Creating a comprehensive evaluation suite is crucial before building, as no single person typically knows all the right answers.

Startups are misapplying the "forward-deployed engineer" (FDE) model. This high-touch, embedded-engineering sales approach is only scalable and justifiable for massive, multi-million dollar contracts like Palantir's, not for typical five-figure startup deals.

For companies building AI agents, the key indicator of a successful customer engagement is the availability of well-documented APIs. These APIs are essential for the agent to take action and look up data, which directly enables a superior, elevated experience from day one.

Unlike other high-risk AI applications, customer service AI can be deployed rapidly in enterprises. The existing infrastructure for escalating issues to human agents provides a natural, low-risk safety net, giving leaders confidence to go live.

Instead of asking for general feedback, Decagon's founder systematized ideation by pressing potential customers on exactly how much they would pay, who approves the budget, and how they would justify ROI. This filters out weak ideas and provides strong commercial signals.

AI's most successful enterprise use cases, customer service and coding, target opposite ends of the labor cost spectrum. It either replaces easily quantifiable, lower-cost roles or provides significant leverage to the most expensive employees like software engineers.

Voice-to-voice AI models promise more natural, low-latency conversations by processing audio directly. However, they are currently impractical for many high-stakes enterprise applications due to a hallucination rate that can be eight times higher than text-based systems.

In rapidly evolving AI markets, founders should prioritize user acquisition and market share over achieving positive unit economics. The core assumption is that underlying model costs will decrease exponentially, making current negative margins an acceptable short-term trade-off for long-term growth.