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

Episode Summary

Executive Summary: Patrick O’Shaughnessy interviews Jesse Zhang, CEO of Decagon, about building AI customer service agents. Zhang explains how Decagon found product-market fit by quantifying customer willingness to pay, why support is a top enterprise AI use case, and how culture, evaluation, voice, data, and distribution shape the next wave of agentic software.

Main Topics: Competitive founder culture (Priority: 5/5): Zhang argues hot markets are inherently competitive, so culture and intensity become durable advantages. Finding product-market fit through pricing signals (Priority: 5/5): Decagon validated ideas by asking buyers exactly what they'd pay and who would approve it. Why customer service is a strong AI wedge (Priority: 5/5): Support has clear ROI, easy rollout, and natural escalation paths that lower enterprise risk. Agents vs coding and labor replacement (Priority: 4/5): Customer service replaces outsourced labor while coding mostly augments highly paid engineers. Operational setup and evaluation (Priority: 4/5): Successful deployments require agreed definitions of quality, tests, guardrails, and feedback loops. Voice, latency, and the enterprise frontier (Priority: 4/5): Voice-to-voice is the next frontier, but hallucination and latency still limit enterprise adoption. Data, moat, and unified brand front end (Priority: 4/5): Conversation data can improve agents over time and eventually become a company's primary interface.

Key Arguments: Competitive markets reward founders with strong culture and execution, not just ideas. Asking how much a customer would pay gives far more signal than asking what they want. Customer service won because ROI is easy to quantify and live rollout is low-risk. Agents can replace outsourced support work, but coding mostly augments expensive engineers. Quality setup is critical: teams must align on what 'good' means before deployment. The best customers are intellectually curious and push hard on AI adoption. Unified agents may become the main front end for brands across support, sales, and retention.

Data Points: Expense review automation: 85% - Ramp claim in sponsor message, using AI to automate expense reviews Expense review accuracy: 99% - Ramp claim in sponsor message Company savings: 5% - Ramp claim in sponsor message: 'Ramp saves companies 5%' Customer willingness to pay: low to mid six figures - Highest amount mentioned in ideation sessions for a potential solution Sleep in New York week: 4 to 5 hours a day - Zhang on his current workload and sleep while traveling Initial rollout size: 5% - Enterprise customers often start with 5% of the user base before expanding One-in-three to one-in-twenty: 1 in 3 to 1 in 20 - Oura Ring example of users pressing agent repeatedly before vs after Decagon Voice hallucination gap: 8x higher - Zhang estimates voice-to-voice hallucinations are about 8x higher than text systems Support interaction share: 90% to 95% voice and 5% chat - Some Fortune 100 customers still operate overwhelmingly over voice Revenue milestone reward: Decagon Arc'teryx jackets - Team milestone incentive used to rally around revenue goals Minimum FDE customer size: probably a million - Zhang says forward-deployed engineers only make sense at roughly million-dollar customer scale

Pivotal Quotes: "There's no challenge that can't be overcome, and there's no enemy that can't be defeated." — Jesse Zhang: The quote on Decagon's office wall that reflects the company's competitive culture "If you really go deep there, it's almost like you're basically asking like classic sales qualification questions, but in founder form." — Jesse Zhang: How Decagon tested ideas by probing customer willingness to pay and approval chains "What we are building towards is this concept of becomes like a new UI for the product." — Jesse Zhang: His vision for agents as the primary front end for brands

Implications: The next battleground is not just model quality but enterprise workflow design, evaluation, and voice reliability; founders should prioritize measurable ROI and customer-specific integration.