A Substack post by Farida Khalaf argues AI models have no moat, using the Clawbot/OpenClaw story as proof. The core thesis — models are interchangeable commodities — is correct. I build on top of LLMs and have swapped models three times with minimal im…