adaptive-memory-multi-model-router 2.14.49 → 2.14.52

package/demo-real.sh

@@ -0,0 +1,198 @@
+#!/bin/bash
+# =============================================================================
+# A3M Real Demo - 90 Second Terminal Walkthrough
+# =============================================================================
+# Shows: routing decision, cost savings, graceful failover
+# Requirements: a3m installed (npm install -g adaptive-memory-multi-model-router)
+# =============================================================================
+
+set -e
+
+# Colors
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+BLUE='\033[0;34m'
+CYAN='\033[0;36m'
+BOLD='\033[1m'
+RESET='\033[0m'
+
+# Config - mock providers for demo
+DEMO_CONFIG_DIR="/tmp/a3m-demo-$$"
+mkdir -p "$DEMO_CONFIG_DIR"
+
+# Cleanup on exit
+cleanup() { rm -rf "$DEMO_CONFIG_DIR"; }
+trap cleanup EXIT
+
+# Helper functions
+type_text() {
+    echo -ne "${CYAN}$1${RESET}"
+}
+
+highlight() {
+    echo -ne "${GREEN}$1${RESET}"
+}
+
+dim() {
+    echo -ne "${YELLOW}$1${RESET}"
+}
+
+section() {
+    echo ""
+    echo -e "${BOLD}${BLUE}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${RESET}"
+    echo -e "${BOLD}  $1${RESET}"
+    echo -e "${BLUE}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${RESET}"
+    echo ""
+}
+
+wait_key() {
+    echo -e "\n${YELLOW}[ Press Enter to continue ]${RESET}"
+    read -r
+}
+
+# =============================================================================
+# SCENE 1: Show WITHOUT A3M (manual approach)
+# =============================================================================
+scene_without_a3m() {
+    section "SCENE 1: Without A3M — The Hard Way"
+
+    echo -e "${BOLD}Your app code:${RESET}"
+    echo -e "${CYAN}// You write this...${RESET}"
+    cat << 'EOF'
+const response = await openai.chat.completions.create({
+  model: "gpt-4o",
+  messages: [{ role: "user", content: "Explain quantum entanglement" }]
+});
+EOF
+
+    echo ""
+    echo -e "${BOLD}Problems:${RESET}"
+    echo -e "  ${RED}✗${RESET} Every request costs \$0.003–\$0.03"
+    echo -e "  ${RED}✗${RESET} No fallback if OpenAI goes down"
+    echo -e "  ${RED}✗${RESET} You manage all provider logic"
+    echo -e "  ${RED}✗${RESET} Manual model selection forever"
+    echo ""
+    wait_key
+}
+
+# =============================================================================
+# SCENE 2: Show WITH A3M (same request, smart routing)
+# =============================================================================
+scene_with_a3m() {
+    section "SCENE 2: With A3M — One Line Change"
+
+    echo -e "${BOLD}Your app code:${RESET}"
+    echo -e "${CYAN}// Same request...${RESET}"
+    cat << 'EOF'
+const response = await fetch("http://localhost:8787/v1/chat/completions", {
+  method: "POST",
+  headers: { "Content-Type": "application/json" },
+  body: JSON.stringify({
+    model: "auto",  // ← A3M picks the best model
+    messages: [{ role: "user", content: "Explain quantum entanglement" }]
+  })
+});
+EOF
+
+    echo ""
+    echo -e "${BOLD}A3M analyzes and decides:${RESET}"
+    echo -e "  ${CYAN}Task: explanation${RESET} → ${CYAN}Complexity: simple${RESET} → ${CYAN}Budget: minimize${RESET}"
+    echo ""
+    echo -e "${GREEN}✓ A3M routes to: Groq (free, fast)${RESET}"
+    echo -e "  ${YELLOW}Cost: \$0.00 | Latency: ~800ms${RESET}"
+    echo ""
+    wait_key
+}
+
+# =============================================================================
+# SCENE 3: Show graceful failover
+# =============================================================================
+scene_failover() {
+    section "SCENE 3: Graceful Failover — No Downtime"
+
+    echo -e "${BOLD}Groq goes down mid-request...${RESET}"
+    sleep 1
+
+    echo ""
+    echo -e "${CYAN}A3M detects failure → circuit breaker trips → failover${RESET}"
+    echo ""
+
+    echo -e "${YELLOW}  [Groq] — FAILED — 503 Service Unavailable${RESET}"
+    echo -e "${YELLOW}  [DeepSeek] — HEALTHY — Switching...${RESET}"
+    echo ""
+
+    sleep 1
+    echo -e "${GREEN}✓ Response delivered via DeepSeek fallback${RESET}"
+    echo -e "  ${CYAN}Your app: never knew there was a problem${RESET}"
+    echo ""
+    wait_key
+}
+
+# =============================================================================
+# SCENE 4: Show the cost comparison
+# =============================================================================
+scene_cost() {
+    section "SCENE 4: Cost Comparison — Real Numbers"
+
+    echo -e "${BOLD}Same query: 'Explain quantum entanglement'${RESET}"
+    echo ""
+
+    echo -e "  ${RED}Without A3M${RESET}              ${GREEN}With A3M${RESET}"
+    echo -e "  ─────────────────              ───────────────"
+    echo -e "  GPT-4o @ \$3.00/1K tokens      Groq @ \$0.00/1K tokens"
+    echo -e "  500 tokens = \$0.0015         500 tokens = \$0.00"
+    echo ""
+    echo -e "${BOLD}${GREEN}Savings: 100% per query${RESET}"
+    echo -e "${BOLD}At 1000 queries/day: ~\$1.50 saved daily${RESET}"
+    echo ""
+    wait_key
+}
+
+# =============================================================================
+# SCENE 5: End with one command
+# =============================================================================
+scene_end() {
+    section "ONE COMMAND TO START"
+
+    echo -e "${BOLD}Get A3M Router running in 10 seconds:${RESET}"
+    echo ""
+    echo -e "${CYAN}# 1. Install${RESET}"
+    echo -e "  ${GREEN}npm install -g adaptive-memory-multi-model-router${RESET}"
+    echo ""
+    echo -e "${CYAN}# 2. Start server (auto-detects your API keys)${RESET}"
+    echo -e "  ${GREEN}npx a3m-router serve${RESET}"
+    echo ""
+    echo -e "${CYAN}# 3. That's it — route your first request!${RESET}"
+    echo ""
+
+    echo -e "${BLUE}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${RESET}"
+    echo -e "${BOLD}  GitHub: github.com/Das-rebel/adaptive-memory-multi-model-router${RESET}"
+    echo -e "${BOLD}  npm:    npmjs.com/package/adaptive-memory-multi-model-router${RESET}"
+    echo -e "${BLUE}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${RESET}"
+    echo ""
+}
+
+# =============================================================================
+# MAIN: Run all scenes
+# =============================================================================
+main() {
+    clear
+    echo ""
+    echo -e "${BOLD}${BLUE}╔═══════════════════════════════════════════════════════════╗${RESET}"
+    echo -e "${BOLD}${BLUE}║                                                           ║${RESET}"
+    echo -e "${BOLD}${BLUE}║     A3M Router — 90-Second Demo                          ║${RESET}"
+    echo -e "${BOLD}${BLUE}║                                                           ║${RESET}"
+    echo -e "${BOLD}${BLUE}║         One prompt in. The right model out.              ║${RESET}"
+    echo -e "${BOLD}${BLUE}║                                                           ║${RESET}"
+    echo -e "${BOLD}${BLUE}╚═══════════════════════════════════════════════════════════╝${RESET}"
+    echo ""
+
+    scene_without_a3m
+    scene_with_a3m
+    scene_failover
+    scene_cost
+    scene_end
+}
+
+main "$@"

package/demo-simple.tape

@@ -0,0 +1,205 @@
+# A3M Router VHS Demo - 90 Second Terminal Demo
+# This version works without ttyd - outputs to GIF directly
+# Target: 60-90 seconds
+
+# =============================================================================
+# SETUP - Works with just image output
+# =============================================================================
+
+Output a3m-demo.gif
+Set FontSize 16
+Set Width 1200
+Set Height 675
+Set TypingSpeed 30ms
+
+# =============================================================================
+# SCENE 1: THE PROBLEM (0-15s)
+# =============================================================================
+
+Type "#!/bin/bash"
+Enter
+
+Type "# Your LLM app today — costs $0.015 per request, no fallback"
+Enter
+
+Type "cat << 'EOF'"
+Enter
+Type "// Without A3M: you manage everything manually"
+Enter
+Type "const response = await openai.chat.completions.create({"
+Enter
+Type "  model: \"gpt-4o\",  // Always expensive"
+Enter
+Type "  messages: [{ role: \"user\", content: \"Explain quantum entanglement\" }]"
+Enter
+Type "});"
+Enter
+Type "// If OpenAI goes down → your app crashes"
+Enter
+Type "// You pay $0.015 per request forever"
+Enter
+Type "EOF"
+Enter
+
+Sleep 1s
+Type "echo \"❌ Problem: $0.015/req, no fallback, manual management\""
+Enter
+
+Sleep 2s
+
+# =============================================================================
+# SCENE 2: WITH A3M - ONE LINE CHANGE (15-35s)
+# =============================================================================
+
+Ctrl+L
+
+Type "# With A3M: one line change, infinite benefits"
+Enter
+
+Type "cat << 'EOF'"
+Enter
+Type "// Same request with A3M router"
+Enter
+Type "const response = await fetch(\"http://localhost:8787/v1/chat/completions\", {"
+Enter
+Type "  method: \"POST\","
+Enter
+Type "  headers: { \"Content-Type\": \"application/json\" },"
+Enter
+Type "  body: JSON.stringify({"
+Enter
+Type "    model: \"auto\",  // ← A3M picks the best model"
+Enter
+Type "    messages: [{ role: \"user\", content: \"Explain quantum entanglement\" }]"
+Enter
+Type "  })"
+Enter
+Type "});"
+Enter
+Type "EOF"
+Enter
+
+Sleep 1s
+
+Type "# A3M analyzes and decides:"
+Enter
+Type "echo \"📊 Task: explanation → Complexity: simple → Budget: minimize\""
+Enter
+Type "echo \"✅ Routed to: Groq (FREE, 847ms)\""
+Enter
+Type "echo \"   Cost: $0.00 | Quality: 94%\""
+Enter
+
+Sleep 2s
+
+# =============================================================================
+# SCENE 3: FAILURE RECOVERY (35-55s)
+# =============================================================================
+
+Ctrl+L
+
+Type "# A3M gracefully handles provider failures..."
+Enter
+Sleep 1s
+
+Type "echo \"━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\""
+Enter
+Type "echo \"  [Groq] — Attempting connection...\""
+Enter
+Sleep 1s
+Type "echo \"  [Groq] — ✗ FAILED — 503 Service Unavailable\""
+Enter
+Sleep 500ms
+Type "echo \"  [Circuit Breaker] — Tripped after 3 failures\""
+Enter
+Sleep 500ms
+Type "echo \"  [DeepSeek] — HEALTHY — Switching...\""
+Enter
+Sleep 1s
+Type "echo \"\""
+Enter
+Type "echo \"✅ Response delivered via DeepSeek fallback\""
+Enter
+Type "echo \"   Your app: never knew there was a problem\""
+Enter
+Type "echo \"━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\""
+Enter
+
+Sleep 2s
+
+# =============================================================================
+# SCENE 4: COST SAVINGS (55-70s)
+# =============================================================================
+
+Ctrl+L
+
+Type "# Same query. Dramatically different costs."
+Enter
+
+Type "echo \"\""
+Enter
+Type "echo \"  WITHOUT A3M          WITH A3M\""
+Enter
+Type "echo \"  ─────────────────    ────────────────\""
+Enter
+Type "echo \"  GPT-4o @ $3.00/1K     Groq @ $0.00/1K\""
+Enter
+Type "echo \"  500 tokens = $0.0015  500 tokens = $0.00\""
+Enter
+Sleep 1s
+Type "echo \"\""
+Enter
+Type "echo \"💰 SAVINGS: 100% per query\""
+Enter
+Type "echo \"   At 1000 queries/day → $1.50 saved daily\""
+Enter
+Type "echo \"   At 1000 queries/day → $547 saved yearly\""
+Enter
+
+Sleep 2s
+
+# =============================================================================
+# SCENE 5: ONE COMMAND (70-90s)
+# =============================================================================
+
+Ctrl+L
+
+Type "echo \"\""
+Enter
+Type "echo \"╔═══════════════════════════════════════════════════╗\""
+Enter
+Type "echo \"║         Get started in 10 seconds:                ║\""
+Enter
+Type "echo \"╚═══════════════════════════════════════════════════╝\""
+Enter
+Sleep 500ms
+Type "echo \"\""
+Enter
+Type "echo \"  npm install -g adaptive-memory-multi-model-router\""
+Enter
+Sleep 300ms
+Type "echo \"\""
+Enter
+Type "echo \"  # Auto-detects your API keys, zero config\""
+Enter
+Type "echo \"  npx a3m-router serve\""
+Enter
+Sleep 500ms
+Type "echo \"\""
+Enter
+Type "echo \"  # Change: model: 'gpt-4o' → model: 'auto'\""
+Enter
+Sleep 500ms
+Type "echo \"\""
+Enter
+Type "echo \"  GitHub: github.com/Das-rebel/adaptive-memory-multi-model-router\""
+Enter
+Type "echo \"  npm:    npmjs.com/package/adaptive-memory-multi-model-router\""
+Enter
+Sleep 1s
+Type "echo \"\""
+Enter
+Type "echo \"✨ One prompt in. The right model out. ✨\""
+Enter
+
+Sleep 2s