Use a Square Menu for Accurate Voice Answers Before Taking Orders
Prepare a selected Square location for menu questions, clarify item options and keep catalog answers distinct from orders or restaurant commitments.
Table of Contents▼
Square AI menu answering can help a caller understand what is listed at a restaurant location before any order is placed. The key is to keep the answer tied to the correct catalog, supported options and current connection rather than letting the assistant fill gaps from general knowledge.
Burki's Square work includes menu search and price quotes, with optional pickup capabilities behind additional setup. Current acceptance evidence is Sandbox-based; live restaurant operation and kitchen fulfillment should not be assumed from those tests.
Select the restaurant location deliberately
A merchant can have more than one location, and menu details may not mean the same thing everywhere. Confirm which location the workspace uses and how staff will maintain the relevant catalog.
The Square setup guide explains the connection and location selection. A caller's mention of another branch should not silently redirect the assistant to a different merchant account or location.
Square's Orders API overview describes its order ecosystem. Burki's menu-answering scope is narrower: finding supported items and options is not itself a completed restaurant order.
Ask enough to identify the item
A caller may use a nickname or omit a size. If several items could fit, present a small, understandable choice. “Did you mean the regular sandwich or the larger meal option?” is better than choosing whichever catalog entry appears first.
Do not invent ingredients, dietary suitability or substitutions because an item name suggests them. If the approved catalog information does not answer a question, explain the gap and direct the caller to the restaurant's actual process.
This is particularly important when a question affects whether someone can safely choose an item. Menu search is not a substitute for staff confirmation of information the restaurant has not supplied.
A hypothetical menu enquiry
Imagine a caller asking whether a pasta dish comes in two sizes and whether a listed side can be added. The assistant finds the relevant item and supported options at the selected location.
It can explain those returned choices and ask which one the caller means before preparing a quote. It should not state that the kitchen has reserved the meal or started preparing it. The caller has asked a menu question, not completed an order.
If the caller changes the size, the quote should use the revised item selection. A previous answer about the smaller portion should not remain attached to the larger one merely because the conversation moved quickly.
Distinguish catalog facts from operating promises
Menu names and configured prices are different from preparation time, live stock or restaurant acceptance. State only the information the supported action returns and the business has approved for use.
A quoted total is an estimate within the configured workflow. It does not reserve stock, take payment or send a kitchen ticket. If the caller asks “So it will be ready when I arrive?” the assistant needs a separate supported outcome before making that promise.
The restaurant voice-assistant page gives the broader use-case context. Keep its conversational possibilities separate from the specific merchant connection and acceptance evidence.
Review the unanswered questions
Ask restaurant staff to list the menu questions that require human judgment: substitutions not represented in the catalog, special preparation, uncertain availability and location-specific exceptions. Those should have explicit fallback wording.
Use clearly synthetic Sandbox items when evaluating setup. Browser practice simulates menu and quote responses, so it does not establish that the production merchant catalog is being read correctly. A working Sandbox quote also does not prove a live caller, payment or fulfillment path.
Before considering customer deployment, verify the exact selected location, representative items, option handling and the behavior when the answer is unavailable. Staff should be able to compare what the assistant said with the authoritative menu information.
Start with one frequently asked menu question and one question the catalog cannot answer. A useful assistant needs to handle both without turning an incomplete answer into a restaurant commitment.
Ready to try Burki?
Create an assistant and check your available browser practice allowance.
Start Free TrialTrial eligibility and available practice are shown in your workspace.