Jesse Zhang, the co-founder and CEO of Decagon, has rapidly emerged as a prominent voice in the world of artificial intelligence, particularly in its application to customer support. His insights, drawn from the experience of being a second-time founder, offer a candid and practical perspective on building a high-growth AI company.
On Building a Company and Entrepreneurship
Quotes:
- "Having a good co-founder makes things way more way more than 2x easier...it's probably like you know order of magnitude like 10x. easier if you have a good one."[1]
- "If you're a founder you're like training your own like neural network right and you need like positive examples and negative. examples."[2]
- "I think the best way to scale is like really getting the leaders right."[3]
- "You can't really overthink the first stage like you can't make it too cerebral... the only way you solve it is by getting as much signal as possible so you just have to go go through the go do the work talk to customers."[3]
- "Everyone knows you have to talk to customers. but even after talking to customers it's really easy to just trick yourself into feeling like oh yeah well I found found an idea. and then you know six months later or like a year later after grinding your butt off you realize oh wait this was a waste of time."[3]
- "The main risk left is execution risk and that's what kind of keeps us on our toes."[1]
- "When you're in the early days...you never really think about like oh Am I competing against other people...for us it was always just talking to customers."[4]
Learnings:
- The immense value of a co-founder: Zhang emphasizes that a strong co-founder is a significant force multiplier, making the challenging journey of a startup substantially more manageable.[1]
- Embrace both successes and failures as learning opportunities: He likens the founder's journey to training a neural network, where both positive and negative outcomes are essential for growth and refinement.[2]
- Prioritize leadership in scaling: According to Zhang, the key to successfully scaling a company is hiring the right leaders.[3]
- Customer-centricity is paramount from day one: One of his most significant lessons is the necessity of being deeply customer-driven from the very beginning to avoid building something nobody wants.[3][4]
- The second-time founder advantage: Zhang suggests that second-time founders often possess a more developed commercial sense and are better equipped to handle the complexities of bringing a product to market.[2]
- Maintain an intense work culture: He believes that a high-intensity work environment is a common denominator among successful AI companies.[2][5]
- The power of in-person collaboration: In the early stages of an AI startup, Zhang advocates for an in-office culture to maximize productivity.[2]
- Strategic market selection is crucial: He underscores the importance of entering the right market at the right time for a startup's success.[6]
- Learn from established companies: Zhang and his team make it a practice to study successful later-stage companies to inform their own growth strategies.[2]
- Navigating international expansion with cultural awareness: He highlights the need to be mindful of cultural nuances when expanding a company globally.[2]
- Product-market fit is the initial, formidable hurdle: The most difficult phase of a startup, in his view, is the initial search for a clear direction and market validation.[7]
On AI and the Future of Technology
Quotes:
- "The promise of an AI agent is like okay well here's a system that can do all the things a human agent can do...it can look up data it can take actions it can write super personalized answers it can review conversations afterwards. and learn things."[4]
- "We see ourselves more or less as a software. company like we do AI. like related products but. I would say most companies now especially application companies you're basically a software. company."[4]
- "If you take AI agent compared to like traditional. SAS. it's almost always a lot easier to measure the impact of an AI agent because you're just comparing it against human labor."[4]
- "For text, the thing that we really care about for models is how good they are at following instructions... On voice, the problem right now is latency."[8]
- "I think it'll probably be mostly anchored in labor costs because that is…that's what is exciting about agents, right, is that you have all this spend in the past that was going towards services."[9]
- "The main innovation of true AI-based customer support solutions is that LLMs are used from the ground up to generate answers, take actions, etc instead of the old way of using decision trees and basic AI models."[10]
Learnings:
- AI agents as a digital workforce: Zhang envisions AI agents evolving far beyond simple chatbots to become a scalable, 24/7 digital workforce capable of complex tasks.[4][9][11]
- Focus on being a software company that uses AI: He makes a clear distinction that Decagon is a software company at its core, utilizing AI as a powerful tool. The defensibility lies in the application layer built on top of foundational models.[4][11]
- The quantifiable ROI of AI: A key advantage of AI agents, according to Zhang, is the ease with which their return on investment can be measured, as it's directly comparable to the cost of human labor.[4][9][11]
- Incremental rollout is key for AI adoption: He advises a gradual implementation of AI agents, starting with specific use cases to prove their value before wider deployment.[8][11]
- Humans and AI will collaborate in customer support: The future of customer service isn't about complete replacement but a partnership where humans supervise, quality-assure, and train their AI counterparts.[8]
- Latency is the primary challenge for voice AI: For voice-based AI to feel natural and be widely adopted, the current issue of response latency must be solved.[8]
- Defensibility lies in customer intimacy and application-level software: In an environment where powerful AI models are becoming commoditized, a company's competitive advantage comes from its deep understanding of customer problems and the software it builds to solve them.[2][5]
- A new, AI-native mindset is required: Zhang argues that the fast-paced evolution of AI demands a more adaptable and user-focused approach than the traditional, more rigid SaaS model.[12]
- Beware of "AI-washing": He cautions buyers to be discerning about whether a product is truly built on modern large language models or is an older system with a superficial AI layer.[10]
On Product and Customers
Quotes:
- "A lot of the value that we've built is in handling logic flows."[8]
- "I can't speak for all companies but customer data is the top priority for us. By default, customer data is never used for training and we have agreements in place with our 3rd-party providers to ensure that they do not train on the data."[10]
- "Our view on this is that in the past, software is based per seat... With most AI agents, the value that you're providing doesn't really scale in terms of the number of people that are maintaining it. It's just like the amount of work output."[9]
- "We've given a lot of thought to this, because we believe that human agents are going to be a core part of how our product gets used."[8]
Learnings:
- Empower non-technical users: A core tenet of Decagon's product philosophy is to enable non-technical users to build and manage AI agents.[3][12]
- Customer data privacy is non-negotiable: Zhang emphasizes a strict commitment to not using customer data for training models by default, building a foundation of trust.[10]
- Pricing based on value delivered: He advocates for a shift away from the traditional per-seat SaaS pricing model to one based on the work output of the AI agent, which more accurately reflects the value provided.[9]
- The choice of AI transparency rests with the customer: Decagon provides its clients with the flexibility to decide whether to disclose to their customers that they are interacting with an AI.[9]
On Personal Growth and the Founder's Mindset
Learnings:
- The transition from short-term to long-term thinking: As a company scales, Zhang notes the critical shift from a short-term focus on immediate goals to a more strategic, long-term perspective.[2]
- The unexpected talent pipeline from the math olympiad community: He observes a fascinating connection between the analytical and problem-solving skills developed in the math olympiad community and success in the AI startup world.[11][13]
- The courage to pivot: Zhang believes that founders should not be afraid to change direction if they lose conviction in their initial idea.[6]
- Aspiring founders should learn from established startups: He advises engineers with entrepreneurial ambitions to join a company that has already achieved commercial success to gain invaluable experience.[2]
- The new wave of intense founders: The current AI landscape, according to Zhang, has attracted a new generation of highly driven and hardworking entrepreneurs.[7]
- Leverage your network for crucial hires: He recommends that founders tap into the expertise of their investors and advisors when making critical leadership hires.[3]
- Rapid learning as a superpower: His journey exemplifies the importance of quickly learning from experiences and applying those lessons to future endeavors.[3][4]
- Don't mistake intelligence for experience: Zhang acknowledges that while young founders may be smart and hardworking, they often lack the business intuition that comes with experience.[5]
- Be wary of benchmarking against unrealistic AI capabilities: He cautions against overpromising what AI can do, for instance, by claiming an "AI agent data scientist" can fully replace a human expert.[11]
Sources
