The session, moderated by Kimberly Tan from Andreessen Horowitz, featured Jesse and Verun, co-founders of Decagon and Clay respectively, two leading enterprise AI companies. Decagon focuses on AI agents for customer support, aiming to transform support from a cost center into a concierge experience, serving clients like Hertz and Duolingo. Clay, on the other hand, is an AI go-to-market platform that helps companies turn growth ideas into reality through data enrichment and automated actions, with customers including HubSpot, Canva, and OpenAI. Both founders shared their journeys, emphasizing how AI has enabled new business models and transformed traditional workflows.
Jesse explained that Decagon’s founding was inspired by the challenges of scaling customer support in a user-heavy consumer company. Through extensive customer discovery, they identified customer service as a problem well-suited for generative AI, allowing measurable ROI and impactful automation. Verun shared that Clay initially started with a broader vision to democratize programming power but found early traction in data enrichment for cold email marketing. Over time, AI advancements allowed Clay to evolve into a platform that supports usage-based pricing and scales go-to-market efforts beyond one-to-one sales.
Both companies highlighted the importance of evaluating new AI models rigorously and maintaining flexible infrastructure. Decagon uses customized evaluation tests and A/B testing to assess new models’ performance, allowing customers to define their own benchmarks. Clay’s architecture separates AI integrations from the core product, enabling rapid iteration and adaptation. They also emphasized the role of user behavior and feedback in guiding product development, citing examples like Decagon’s web research agent and conversational AI co-pilot.
Addressing enterprise concerns about AI safety and governance, both founders stressed the need for customizable guardrails and user empowerment. Decagon introduced “Agent Operating Procedures” (AOPs), which allow non-technical users to set and manage AI behavior rules, reducing risk and increasing trust. They also noted a shift in enterprise mindset from fearing isolated AI errors to focusing on overall error rates, which should be lower than human error rates. Both companies advocate for gradual rollouts with human-in-the-loop systems to ensure reliability and build confidence.
When discussing market differentiation and growth strategies, Jesse and Verun emphasized winning through product excellence and unique philosophies rather than competing on noise alone. Clay differentiates by leveraging proprietary data aggregation and AI-driven workflows, while Decagon focuses on democratizing AI configuration for non-technical users. Both companies prefer horizontal market approaches, serving multiple industries with adaptable solutions rather than deep vertical specialization. Finally, they advised aspiring enterprise AI founders to focus on introspection, curiosity, and iterative learning rather than over-planning or blindly following external advice.
