YOu don’t use function calling from the front end. The idea behind function calling is that the openAI conversation is handled by the “tool runner”, which is some kind of server-side resource that knows how to answer the requests for tool executions. YOu can do this in no-code using one of the backends out there (Xano, Fastgen, Buildship, Comnoco) or more code-y backend alternatives (Supabase, lambdas, etc)