2nd/10_Wiki/Topics/AI_and_ML/Point-of-Sale.md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-point-of-sale	Point of Sale	10_Wiki/Topics	verified	self	POS retail-pos payment-terminal	none	A	0.9	applied	retail payments pos square stripe		2026-05-10	pending	language framework TypeScript Stripe Terminal / Square API

Point of Sale (POS)

매 한 줄

"매 retail transaction 의 capture system — 매 hardware + software + payments + analytics". 매 cash register → ECR (1970s) → cloud POS (Square 2009, Toast 2011, Shopify POS) → 매 modern unified commerce (online + offline + AI inventory + loss prevention).

매 핵심

매 components

• 매 hardware: terminal (iPad/Android), card reader (chip + tap + Apple Pay), receipt printer, cash drawer, barcode scanner, kitchen display.
• 매 software: catalog, cart, tax, discounts, tips, split bill, refunds, customer profiles, inventory.
• 매 payments: card-present (EMV chip + NFC), digital wallets (Apple Pay, Google Pay), QR (Alipay, WeChat), BNPL (Klarna, Afterpay).
• 매 analytics: sales dashboards, item performance, employee performance, demand forecast.
• 매 integrations: accounting (QuickBooks), payroll, e-commerce (Shopify), marketing (loyalty, SMS).

매 modern players (2026)

• 매 Square (Block): SMB-focused, Cash App integration, free entry POS.
• 매 Stripe Terminal: developer-first, omnichannel API, embedded readers.
• 매 Toast: restaurant-vertical, kitchen ops, payroll.
• 매 Shopify POS: omni-commerce (online + retail unified inventory).
• 매 Lightspeed: retail/restaurant/golf vertical.
• 매 Clover (Fiserv): bank-distributed, app marketplace.
• 매 SumUp / iZettle: Europe SMB.

매 AI/ML in POS (2026)

• 매 loss prevention: computer vision 매 self-checkout 매 mis-scans (Everseen, Diebold, Mashgin Vision).
• 매 demand forecasting: per-SKU per-store sales prediction (XGBoost/transformer).
• 매 dynamic pricing: 매 perishables markdown optimization.
• 매 fraud detection: 매 chargeback prediction, 매 stolen card flagging.
• 매 voice ordering / AI drive-thru: McDonald's, Wendy's deployments.
• 매 visual search: scan product 매 photo 의 lookup.

매 standards / compliance

• 매 PCI-DSS: card data handling.
• 매 EMV: chip card protocol.
• 매 P2PE: point-to-point encryption.
• 매 tokenization: store token, not PAN.
• 매 SOC 2: SaaS POS providers.

매 응용

1. 매 retail (apparel, grocery, c-store).
2. 매 restaurants (QSR, FSR, food trucks).
3. 매 services (salons, mobile contractors).
4. 매 events / pop-ups (mobile readers).

💻 패턴

Stripe Terminal — collect in-person payment

import { loadStripeTerminal } from "@stripe/terminal-js";

const StripeTerminal = await loadStripeTerminal();
const terminal = StripeTerminal.create({
  onFetchConnectionToken: async () => {
    const r = await fetch("/connection_token", { method: "POST" });
    return (await r.json()).secret;
  },
  onUnexpectedReaderDisconnect: () => console.warn("disconnected"),
});

const discoverResult = await terminal.discoverReaders();
await terminal.connectReader(discoverResult.discoveredReaders[0]);

// Server creates PaymentIntent, client collects
const { paymentIntent } = await fetch("/create_pi", {
  method: "POST", body: JSON.stringify({ amount: 1999 }),
}).then(r => r.json());

const result = await terminal.collectPaymentMethod(paymentIntent.client_secret);
const confirmed = await terminal.processPayment(result.paymentIntent);
console.log("paid:", confirmed.paymentIntent.id);

Square Terminal API

const { Client, Environment } = require("square");
const client = new Client({
  accessToken: process.env.SQUARE_TOKEN,
  environment: Environment.Production,
});

const checkout = await client.terminalApi.createTerminalCheckout({
  idempotencyKey: crypto.randomUUID(),
  checkout: {
    amountMoney: { amount: 1999n, currency: "USD" },
    deviceOptions: { deviceId: "DEVICE_ID" },
  },
});

Catalog + tax calculation

type Item = { id: string; name: string; price_cents: number; tax_rate: number; sku: string };

function calculateOrder(items: Item[], qty: Record<string, number>, discount_cents = 0) {
  const subtotal = items.reduce((s, it) => s + it.price_cents * (qty[it.id] ?? 0), 0);
  const tax = items.reduce((s, it) =>
    s + it.price_cents * (qty[it.id] ?? 0) * it.tax_rate, 0);
  const total = subtotal + tax - discount_cents;
  return { subtotal, tax, discount: discount_cents, total };
}

Receipt printer (ESC/POS)

// node-thermal-printer
const { ThermalPrinter, PrinterTypes } = require("node-thermal-printer");
const printer = new ThermalPrinter({
  type: PrinterTypes.EPSON,
  interface: "tcp://192.168.1.100",
});
printer.alignCenter(); printer.println("ACME COFFEE");
printer.drawLine();
printer.tableCustom([
  { text: "Latte x2", align: "LEFT" },
  { text: "$10.00", align: "RIGHT" },
]);
printer.drawLine();
printer.println("TOTAL: $10.80");
printer.cut(); await printer.execute();

EMV / NFC reader (BBPOS via Stripe)

// 매 Stripe Terminal SDK 매 BBPOS WisePOS E / Stripe Reader S700
// 매 PCI-DSS scope 의 reduce — 매 reader 매 PAN 의 handle, 매 server 매 token 의 receive
const reader = await terminal.connectReader(readers[0], { fail_if_in_use: true });

Inventory sync (Shopify POS)

// Shopify Admin API — unified inventory across channels
const res = await fetch(`https://${SHOP}.myshopify.com/admin/api/2025-01/inventory_levels/adjust.json`, {
  method: "POST",
  headers: { "X-Shopify-Access-Token": TOKEN, "Content-Type": "application/json" },
  body: JSON.stringify({
    inventory_item_id: 12345,
    location_id: 67890,
    available_adjustment: -1,  // sold one
  }),
});

Loss prevention (CV self-checkout)

# 매 simplified pseudocode — 매 actual: YOLOv8 + tracking + match to scanned barcode
import cv2
from ultralytics import YOLO

model = YOLO("yolov8n.pt")
cap = cv2.VideoCapture(0)
scanned_skus = set()  # populated by barcode scanner events

while True:
    ret, frame = cap.read()
    detections = model(frame)[0]
    visible_items = {classify_to_sku(d) for d in detections.boxes}
    discrepancy = visible_items - scanned_skus
    if discrepancy:
        alert_associate(f"Unscanned: {discrepancy}")

Webhook — payment.succeeded

// 매 Stripe webhook 의 receive — 매 settle order in DB
import express from "express";
const app = express();
app.post("/webhook", express.raw({ type: "application/json" }), async (req, res) => {
  const sig = req.headers["stripe-signature"]!;
  const event = stripe.webhooks.constructEvent(req.body, sig, WEBHOOK_SECRET);
  if (event.type === "payment_intent.succeeded") {
    const pi = event.data.object;
    await db.orders.update({ stripe_pi: pi.id }, { status: "paid" });
  }
  res.json({ received: true });
});

매 결정 기준

상황	Solution
매 SMB cafe / pop-up	매 Square
매 restaurant chain	매 Toast
매 omni-commerce retail	매 Shopify POS
매 dev-first / custom UX	매 Stripe Terminal
매 enterprise multi-store	매 NCR / Oracle Symphony
매 EU SMB	매 SumUp / iZettle
매 unattended kiosk / self-checkout	매 Mashgin / Toshiba + Stripe

기본값: 매 Stripe Terminal (developer-first, custom flow) 또는 매 Square (turnkey SMB).

🔗 Graph

• Adjacent: Shopify

🤖 LLM 활용

언제: 매 receipt parsing, 매 customer Q&A chatbot, 매 menu generation, 매 anomaly summary 의 사용. 언제 X: 매 payment authorization (deterministic flow), 매 PII handling (compliance risk).

❌ 안티패턴

• 매 store full PAN: 매 PCI-DSS violation. 매 token 만 store.
• 매 home-grown payment terminal: 매 EMV cert 매 expensive — 매 SaaS provider 의 사용.
• 매 ignore offline mode: 매 retail Wi-Fi 매 unstable — 매 store-and-forward needed.
• 매 monolithic POS: 매 vertical lock-in — 매 API-first (Stripe/Shopify) 의 prefer.
• 매 no idempotency keys: 매 double-charge risk on retry.

🧪 검증 / 중복

• Verified (Stripe Terminal docs, Square API, Shopify Admin API, PCI-DSS v4.0).
• 신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — Stripe/Square/Shopify patterns + AI loss prevention