As of 13 August 2026, AI can only partly tell customers whether a product is in stock.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Hand it to a person

   Second choice

   A person who owns the outcome does this end to end, worth it when the failure is dear.

3. Do it yourself

   The distant third

   A chat interface, developer skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open the stock or order-management system and list the fields needed to identify an item, including SKU, product name, variant, quantity and shop or warehouse location.
   2. Ask the system owner for the approved API, webhook or other live inventory connection, and record what its available, unavailable, stale and error responses mean.
   3. Paste the product fields, stock rules and connection details into the prompt, replacing each bracketed slot with your business information.
   4. Ask the AI to produce the response rules and test conversations, then configure the chosen chatbot or support platform to call the live inventory lookup before it replies.
   5. Run the five test conversations using known products and compare every drafted answer with the current stock records, including variant and location details.
   6. Send the flow to a colleague who handles orders, have them test ambiguous names and failed lookups, and change the fallback message until it never presents unconfirmed availability as fact.
   7. Publish the chatbot only after the live lookup, error fallback and human contact route work in the customer channel where stock questions arrive.

   Prompt

   Design a customer-support flow that tells customers whether a product is in stock using the live inventory source described below.
   
   Business context:
   - Business: [business name]
   - Sales channels: [website, shop, phone or other channels]
   - Stock system: [system name]
   - Inventory connection available: [API, webhook, database query or none]
   - Product fields available: [SKU, product name, variant, quantity, warehouse, shop location and any other fields]
   - Stock rules: [for example, available only when quantity is above zero, or available for collection only at a named shop]
   - Customer questions to support: [examples]
   
   Produce:
   1. The exact answer format the AI should use for available, unavailable, uncertain and discontinued products.
   2. The fields it must ask for when a product name or variant is ambiguous.
   3. A safe workflow for looking up the live stock record before answering.
   4. A fallback message that says it cannot confirm availability when the connection is missing, stale or returns an error.
   5. Five test conversations covering an available product, an unavailable product, an ambiguous variant, a shop-specific query and a failed inventory lookup.
   
   Never claim that an item is in stock from static product content alone. Never invent a quantity, location, delivery time or restock date. If the live result is missing or unclear, say that availability cannot be confirmed and offer the approved human contact route.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.