No PriorsNo Priors Ep. 132 | With Decagon CEO and Co-Founder Jesse Zhang
CHAPTERS
- 0:00 – 1:11
Meet Jesse Zhang & Decagon’s AI concierge for enterprises
Elad introduces Jesse Zhang, CEO/co-founder of Decagon, and sets the stage for how Decagon uses AI agents to handle high-volume customer interactions for major enterprises. Jesse explains Decagon as a conversational “concierge” interface for a brand—designed to resolve issues, reduce costs, and ultimately improve engagement.
- •Decagon builds AI customer service agents for banks, airlines, telcos, and major tech companies
- •Positioning: personalized, engaging conversations that resolve issues end-to-end
- •Long-term framing: conversational UI/concierge becomes a primary way users interact with brands
- 1:11 – 2:41
From digital natives to big banks: how Decagon landed large customers fast
Jesse describes starting with digital-native companies that adopt quickly, then being pulled upmarket by demand and contact volume. They discuss how the AI platform shift has created unusually top-down urgency inside enterprises, accelerating buying cycles.
- •Initial traction with digital natives helped iterate quickly (e.g., Notion)
- •Enterprises moved faster than expected due to AI transformation pressure
- •AI adoption is increasingly top-down (C-suite/board driven)
- •Customer service is viewed as low-hanging fruit for AI ROI
- 2:41 – 3:33
Measuring ROI: efficiency gains and customer satisfaction outcomes
The conversation turns to how enterprises evaluate Decagon’s impact. Jesse highlights the primary metric of contact-center cost reduction, alongside the equally important requirement to maintain or improve customer satisfaction.
- •Primary enterprise KPI: efficiency/cost reduction in contact center operations
- •Case studies cite major cost reductions (e.g., 60–70% in strong outcomes)
- •Secondary (often equal) KPI: customer satisfaction and engagement
- •Clear business cases help internal champions justify deployments
- 3:33 – 4:25
Integration into existing workflows: agents as labor substitutes, not tool replacements
Jesse explains Decagon’s deployment philosophy: integrate with the customer’s existing CRM, telephony, and support stack rather than disrupting it. He emphasizes the operational advantages of AI agents—always on, scalable, and without human training/churn constraints.
- •Agents integrate into existing CRMs and telephony stacks
- •Framed as substituting mundane labor while preserving current tooling
- •Operational benefits: 24/7 availability, no churn, rapid scaling
- •Over time, deployments expand as agent capability and trust grow
- 4:25 – 5:40
Second-time founder lessons: becoming more commercially disciplined
Elad and Jesse discuss how second-time founders often bring more go-to-market and customer focus. Jesse shares a thesis: strong technical talent can be “trained” to be more commercial, and that the messy GTM problems can be intellectually rewarding and key to speed.
- •Thesis: technical people can become highly effective commercially
- •GTM is messy but still problem-solving—and critical to growth
- •Second-time experience improves intuition on what’s a good idea
- •Founder partnership: shared technical background and stage-of-life alignment
- 5:40 – 8:32
Hiring philosophy and culture: optimize for intelligence and intensity (in-office)
Jesse outlines Decagon’s hiring approach: prioritize raw intelligence over narrow experience, while balancing seniority early on. They also discuss Decagon’s five-days-in-office culture, emphasizing productivity, craft, and momentum in early-stage AI companies.
- •Top hiring filter: “smart people” over direct prior experience
- •Early team leaned experienced; later added new grads to the mix
- •Applies across functions (engineering, sales, marketing)
- •Five days in-office; weekends optional—focus on high-output environment
- •Belief: in-office matters more at early stage for AI startups
- 8:32 – 9:13
Scaling from ~50 to ~200: adding structure, leaders, and a real people function
As Decagon approaches ~200 employees, Jesse describes shifting to “building for scale.” That includes more IC hiring, layering in leadership, and investing in org design and operating cadence—plus the complexities of multi-office growth.
- •Hiring for scale: more ICs plus additional leadership layers
- •New priority: dedicated people function (org design, operating cadence)
- •Expanding footprint: New York office and plans for Europe
- •Recognizing the operational overhead that arrives with size
- 9:13 – 15:37
Founder advice during hypergrowth: shift from short-term wins to medium/long-term bets
Jesse explains the mental shift required after early PMF: you must invest in work that won’t immediately close customers but prevents future bottlenecks. He also notes learning by studying later-stage companies that executed scaling well (e.g., Ramp, Databricks).
- •Early-stage: “greedy” short-term focus is rational (close deals, land logos)
- •Post-PMF: obligation to plan longer-term or systems break later
- •Invest in core product/platform work that reduces future per-customer overhead
- •Study successful scalers (Ramp, Databricks) to borrow operating patterns
- 15:37 – 16:57
Why customer service: discovering the wedge through customer willingness to pay
Jesse shares they didn’t start with a fixed plan to do customer service—rather, they ran a disciplined customer discovery process. The decisive signal was strong willingness to pay meaningful contracts even when the company was just two people.
- •No preconceived notion—problem selection emerged from customer discovery
- •Empathy from first company’s large user base informed interest
- •Key validation: customers willing to sign six-figure contracts early
- •“Obvious” ideas still hide nuance and complexity once you build them
- 16:57 – 19:39
AI labs pushing into apps: what that means for Decagon and enterprise defensibility
Elad raises the pattern of platform providers forward-integrating into applications, and asks about foundation model labs moving into verticals. Jesse argues labs will start with consumer/prosumer apps, while enterprise deployments require a thick software layer beyond models.
- •Labs have incentives to move up the stack where margins and customer ownership are stronger
- •Consumer/prosumer verticals likely come first due to simpler packaging
- •Enterprise requires extensive scaffolding: monitoring, insights, QA, simulation/testing
- •Decagon focuses on building that “thick layer” rather than over-forecasting labs
- •Coding seen as a likely priority vertical for labs
- 19:39 – 22:05
What Decagon does uniquely: productized AI agents built for non-technical operators
Jesse describes Decagon’s differentiation as execution speed plus a strongly productized approach. A core bet is empowering non-technical business users to build, iterate, and analyze agents—reducing reliance on slow, expensive configuration-heavy SaaS paradigms.
- •Differentiators: speed of execution and a young, intense team culture
- •Productized approach: easy for non-technical users to iterate on agents
- •Contrast with traditional SaaS (e.g., heavy configuration requiring specialists)
- •Hybrid model: engineering teams handle integrations; business users own logic/iteration
- •Belief: this is the right operating model for the AI era
- 22:05 – 24:46
Pricing services in the AI age: per-conversation economics and expanding TAM
They discuss the shift from SaaS seat pricing to “labor/cognition as a service.” Jesse explains why customer service is naturally priced per conversation (clear measurable output), and how that reframes TAM to include services spend, not just software seats.
- •Pricing depends on use case; for customer service, output = conversation/contact
- •Per-seat doesn’t fit; per-minute can create bad incentives (long calls)
- •Common model: customers buy an allotment of conversations and burn it down
- •TAM expands from software budgets to services spend (contact center economics)
- •Industry remains early: current agent vendors are tiny relative to total market
- 24:46 – 25:52
Beyond reactive support: the “concierge” across the full customer journey
Elad points out customer service starts at first website visit, not just post-purchase issues. Jesse explains Decagon’s “concierge” framing—and the organizational reality that different conversation types map to different internal teams and budgets, even if leaders want a unified end-user experience.
- •Concierge vision spans pre-sales, support, and broader brand interaction
- •Initial assumption: one agent for “any conversation”; reality: org silos and budgets
- •Leaders want unified experiences even if internal ownership is fragmented
- •Long-term: users may prefer interacting with an agent over apps/websites
- 25:52 – 27:41
Defining long-term success: powering major brands with sharp product and execution
Jesse defines success as becoming the category winner powering conversations across major brands and reinventing how consumers interact with companies. He also emphasizes building an enduring company known for thoughtful execution—taking inspiration from elite operators like Databricks and Ramp.
- •Goal: work with the largest companies and become the market winner
- •Vision: reinvent consumer-brand interaction through AI conversations
- •Operational ambition: be “sharp” in product and GTM execution
- •Avoid over-planning decades out; focus on compounding step-by-step progress
- 27:41 – 31:22
An agentic future: agents talking to agents, proactive support, and assisted commerce
Jesse predicts agentic interactions are already emerging: personal agents will increasingly transact and negotiate on users’ behalf, sometimes directly with enterprise support agents. Near-term communication stays natural-language for human compatibility, but over time agent-to-agent protocols may become more efficient and proactive.
- •Personal agents will handle tasks like rebooking, shopping, and resolving issues
- •Near-term: natural language persists because humans remain in the loop
- •Longer-term: more efficient agent-to-agent communication may develop
- •Support expands from reactive issue resolution to proactive outreach and upsell
- •Future state: interactions happen “outside humans” while outcomes still complete
