diff --git a/Makefile b/Makefile
index a44f4ae..8416f23 100644
--- a/Makefile
+++ b/Makefile
@@ -195,7 +195,7 @@ friction: c-build asm-build
 
 GPU_WORKER_DIR := examples/cuda-fanout
 PORT ?= 9091
-LUMBDA ?= python
+LUMBDA ?= c
 
 gpu-worker-bin:
 	$(MAKE) -C $(GPU_WORKER_DIR) shake256-fanout
@@ -203,14 +203,16 @@ gpu-worker-bin:
 gpu-worker: gpu-worker-bin
 	@echo "→ launching gpu-worker.lsp on port $(PORT) ($(LUMBDA) tier)"
 	@printf '(load "wire.lsp")\n(load "gpu-worker.lsp")\n(main)\n' > $(GPU_WORKER_DIR)/launch.lsp
-ifeq ($(LUMBDA),python)
-	cd $(GPU_WORKER_DIR) && python3 -u ../../lumbda.py launch.lsp --port $(PORT)
-else ifeq ($(LUMBDA),c)
+ifeq ($(LUMBDA),c)
+	$(MAKE) c-build
 	cd $(GPU_WORKER_DIR) && ../../c/lumbda launch.lsp --port $(PORT)
+else ifeq ($(LUMBDA),python)
+	cd $(GPU_WORKER_DIR) && python3 -u ../../lumbda.py launch.lsp --port $(PORT)
 else ifeq ($(LUMBDA),asm)
+	$(MAKE) asm-build
 	cd $(GPU_WORKER_DIR) && ../../asm/lumbda-gc launch.lsp --port $(PORT)
 else
-	@echo "LUMBDA=$(LUMBDA) — expected python | c | asm"; exit 2
+	@echo "LUMBDA=$(LUMBDA) — expected c | python | asm"; exit 2
 endif
 
 gpu-worker-test: gpu-worker-bin
diff --git a/www/index.html b/www/index.html
index a79fb21..4f68675 100644
--- a/www/index.html
+++ b/www/index.html
@@ -67,11 +67,12 @@ make test-all</code></pre>
 <pre><code># On any host with nvcc + a CUDA-capable GPU:
 make gpu-worker
 # → builds examples/cuda-fanout/shake256-fanout
-# → launches gpu-worker.lsp on port 9091 (Python tier)
+# → builds the C tier (~10× faster wire orchestration than Python)
+# → launches gpu-worker.lsp on port 9091
 
 # Override tier or port:
-make gpu-worker LUMBDA=c PORT=9001
-make gpu-worker LUMBDA=asm</code></pre>
+make gpu-worker LUMBDA=python PORT=9001   # easier debugging
+make gpu-worker LUMBDA=asm                # smallest footprint</code></pre>
 
 <h3>Call it from any tier</h3>
 <pre><code>;; bend works on every tier — Python, C, asm — through the same