You are helping me build the following AI agent workflow. ## Goal Restaurant POS Orders and Inventory: Captures POS orders, processes payments, updates stock levels, and logs sales to Google Sheets in real time. ## Specification - What it does: Streamline your restaurant operations by automatically capturing orders, processing payments, and updating stock levels without manual entry. You get real-time sales reports in Google Sheets and accurate inventory tracking that prevents overselling while you focus on serving customers. - Trigger: Triggered by an event (Event ยท on new order from POS) - Autonomy: Fully hands-off - Expected setup effort: a few focused days - Tools/services involved: - POS System: retrieve order details and payment info - Payment Gateway: process transactions securely - Inventory Management Tool: update stock levels automatically - Google Sheets: generate real-time sales reports - OpenAI Chat Model: extract and structure order data ## Known pitfalls, handle each one explicitly in your implementation 1. Currency formatting differences between regions can break parsing, normalize all monetary values to a single format before sending them to the LLM. 2. Inventory updates might race if multiple orders arrive simultaneously, implement a queue or locking mechanism to prevent overselling. ## Reference implementation https://n8n.io/workflows/3942 (workflow template) Fetch and inspect this before building. If it matches my stack, adapt it; if not, rebuild the pattern with my tools. ## Process requirements 1. Before building: ask me which of the listed tools I actually use and what my platform is (n8n / Make / code / other). Do not assume. 2. Adapt the pattern to my answers; do not force the reference stack. 3. Address every pitfall above; tell me how you handled each. 4. Provide a test plan I can run before letting this touch real data. 5. Ask before any step that sends messages, modifies data, or spends money. Source: https://usecasesforagents.com/use-case/automate-restaurant-pos-orders-and-inventory/ via usecasesforagents.com