
Ryan Wang, Assembled
SaaS Pricing Trap: Usage-Based Models Need Minimums to Survive
Ryan Wang is the co-founder and CEO of Assembled, an AI platform for customer support that helps companies manage both human and AI agents more efficiently. In 2016, Ryan was a machine learning engineer at Stripe. He and his co-founders spent two years building before launching in 2020—the same day WHO declared COVID a global pandemic. Their momentum vanished. About a quarter of demos didn't show up. Their SaaS pricing model—usage-based with no minimums—meant customers could scale to zero without leaving. It took 8 months to earn their first dollar of revenue. In 2016, Ryan was a machine learning engineer at Stripe. He and his future co-founder Brian built ML tools to automate support tickets, but they realized the real problem wasn't automation—it was workforce management. That became the spark for Assembled. The three co-founders spent two years building before they launched in 2020. They lined up a TechCrunch story, hit the front page of Hacker News, and then their launch landed the same day the World Health Organization declared COVID a global pandemic. Momentum vanished. About a quarter of demos didn't show up. It took them eight months to earn their first dollar of revenue. The SaaS pricing trap: When they finally got customers, they had usage-based pricing with no minimums. Customers could scale usage to zero. When usage flatlined during the pandemic, the team blamed themselves before realizing customers weren't leaving because of the product—they were just cutting costs. How Ryan fixed the SaaS pricing problem: 1. Shifted focus from chasing growth to serving customers who were getting value 2. Met customers in person, sat with support leaders, and built what actually mattered 3. Added pricing minimums to prevent revenue from dropping to zero 4. Built sticky features that justified the investment That hands-on approach worked for about 10 customers. Then it broke at 50. Onboarding took weeks. Some features worked in demos but failed in production. So they rebuilt onboarding to get it down to days and cleaned up the product so it could scale. Eventually they grew from their early customers to dozens more and reached 8-figure ARR.


