Matt Murphy of Menlo Ventures shares insights from investing in Anthropic, emphasizing that a great model alone isn't enough for success—you need to build a 'harness' around it to make it accessible and easy to use. He also discusses the multi-model future of AI, the need for sensible regulation, and the unprecedented growth speeds of AI startups, which are product-led and have low adoption friction.
Summarized by Podsumo
Anthropic's growth from $0 to $47B revenue run rate in three years is unprecedented.
The key lesson for founders: a great model (tech) is not enough; you need a 'harness' (MCP, skills, data connectivity) to make it accessible.
AI is product-led (not sales-led), with low adoption friction, enabling startups to grow from zero to $300M in a year.
Murphy predicts a multi-model future: enterprises will use proprietary (Claude), open-source, and custom models based on cost, latency, and data needs.
The biggest next frontier is consumer AI agents that help individuals fight companies (e.g., insurance claims, lower prices).
"_"The lesson for founders now is that a great model isn't enough. You need to build a harness around it to make it accessible and easy to use."_ — Matt Murphy"
"_"The hard part was investing at a $4 billion valuation pre-revenue, pre-launch. That was a tricky part."_ — Matt Murphy"
"_"The consumer thing won't be me, I'm afraid. All going to be possible."_ — Matt Murphy"