Decagon

It was a pleasure chatting with Jesse Zhang from Decagon on the future of AI agents and applied AI. We talk about * how AI is reshaping customer support * how Decagon thinks about building differentiated products * the evolution of software pricing * the impact of AI agents on the enterprise Listen on Spotify for the AI + Andreessen Horowitz podcast at the link below!

Our engineering team is growing! Huge welcome to: Stefan Grasu from Heasy and Ladder 🥳 Allen Gu from Scale AI and University of California, Berkeley 🎉 (Thanks to Cynthia Chen for the referral!) Christopher Chen from Moveworks and Roblox 🎊 It's a super exciting time for the team, and we're still aggressively hiring (in-person in SF). If you have any recommendations for engineering, sales, or marketing... send them over!

There's been a lot of buzz recently in the tech community around how AI agents should be priced (check out Aaron Levie's thoughtful post). Today, we're sharing some of our learnings around the economics of deploying AI agents to many customers. At the core, the value provided by AI agents is fundamentally different from the traditional per-seat software products we know and love. AI agents are able to complete entire jobs autonomously, so there's a direct parallel with human labor. Similarly, there's often an easy way to quantify and measure the impact. In our space at Decagon, the metrics everyone evaluates are: 1. Deflection rate: What % of customer conversations are resolved by the AI agent with needed a human? 2. Customer satisfaction: How happy are customers with these conversations? This naturally leads to a pricing model that scales with conversation volume, because that's how much work is done. We offer 2 ways to achieve this: 1. Per-resolution: Pay $X for each conversation that's resolved by the AI 2. Per-conversation: Pay $X for each conversation that flows through the AI Over time, we've seen the vast majority of customers opt for the per-conversation model. Why is that? Per-resolution has the benefit of scaling with the 'results' achieved by the AI. At the same time, it naturally leads to a debate: what exactly is a 'resolution'? You never want to be in a situation where a user is upset and leaves, but they're just being deflected because that's how the incentives work. We've seen that companies prefer the predictability and collaborative-ness of the per-conversation model, but both options have merit. I'm excited to see how the pricing models play out in other AI agents well! For example, I've heard of some creative models on the coding side. Full post linked below by Bihan Jiang! 👇

Money never sleeps. Support shouldn't either. Finance is one of those industries where support *should* be really, really good — but that's not the reality today. Customers constantly face friction between limited 9 am - 5 pm support hours, hours-long wait times, and unhelpful self-service portals. This is really problematic because concerns involving money are often urgent, like suspected fraud or failed payments. 👉 At Decagon, we're not just fixing broken fintech support — we're transforming customer relationships. By delivering experiences that actually delight, we're building the foundation for truly customer-centric financial services. Our gen-AI agents deliver the immediate, 24/7 support that modern customers deserve. Learn more about specific fintech use cases in our blog post linked in the comments 🔗

"Are you gonna put that on LinkedIn?" Of course. We love Decagon smoothie days 🫶

Still buzzing from an incredible week at #AWSreinvent! 🚀 In a year defined by generative AI breakthroughs, Jesse Zhang and Adara Parker delved into many insightful conversations about the future of AI agents and customer experience. Some highlights 👇 - Deep dives into building resilient, scalable AI agent architectures 📈 - Coffee chats with Heads of GenAI and CX executives 💫 - Thought-provoking dinners hosted by Accel, Andreessen Horowitz, Amazon Web Services (AWS), Baseten, and Orb 🍽️ Feeling energized from all our learnings and looking forward to continuing conversations. Back to building! 🏃 #AWS #Cloud #AI #Innovation #FutureOfAIAgents

Enterprise AI should be transparent AI. Read more on our blog 🔗👇

🌟 Engineer Spotlight: Meet Viraj! We’re kicking off our spotlight series for our exceptional team at Decagon. Today, we're excited to highlight Viraj Maddur, one of our wonderful software engineers! Viraj is a UT Austin alum (#hookem), where he became a proficient user of “y’all” and won several algorithmic programming tournaments, qualifying his team for the ICPC World Finals in Moscow in the process. When he's not making PRs, you might find Viraj: 🐕 part-time dog walking, 🥁 learning the drums (shoutout Zac Farro), or 🃏 playing no-limit hold ‘em poker. Viraj decided to join Decagon because "Everyone wants to win and the energy is incomparable because of it." Let's goooo 👏 That competitive spirit showed when his team won our Drunk AIME competition (when he can Lagrange multiply while sipping on a Nonagon Negroni 😍). We’re so grateful to have Viraj in our engineering family! #EmployeeSpotlight #DecaFam #Engineer #CompetitiveProgramming

Excited to see Decagon as one of the leading companies at the agentic layer!

🎉Thrilled to welcome Ryan Buick to Decagon! Ryan joins us from Canvas (co-founder & Head of GTM) with prior expertise from Flexport and AppDirect. His strategic vision and proven track record in building successful market strategies will be invaluable as we scale. Welcome aboard, Ryan! We're excited for the impact you'll create here.