Keep the restaurant demo separate from real guest service

The decision this workflow supports

A prospect may want to hear Ava answer as their own restaurant. A demo should use a small fictional business profile and make clear that it is demonstrating request handling rather than connecting to the restaurant’s real systems.

Put the review process in place

Ask for public or fictional business details only, then show the profile being demonstrated. Keep the session isolated and label any resulting intake as a sample. Do not request guest records, platform passwords, payment details or real reservation data to make the demonstration feel realistic.

Worked example for your team

Open two fictional demonstrations in separate sessions: one for a neighborhood dining room and one for an office-catering business. Ask the same question about a future event in both. Each response should use only the profile shown in its own session. A sample intake must remain labeled demonstration even if it resembles a real staff note. Then close one session and start another with different details. The new visitor should not inherit the earlier profile or transcript. This checks isolation and honest presentation without connecting any real restaurant booking or payment system.

Sample wording to approve

“This is a demonstration using the profile shown here. Please use fictional guest and event details. Any intake or follow-up you see is a sample, not a restaurant booking or order.” This wording is a draft for the restaurant owner to approve before use.

Fields for the staff handoff

Staff need demo business name, fictional service details, scenario selected, sample result label, and session owner. Mark uncertain details explicitly and keep the requested action separate from anything restaurant staff have already confirmed.

Acceptance tests before use

Two visitors do not see each other’s profiles. A sample request is clearly labeled. The demonstration cannot create a real booking, order or payment. Record the observed result, assign any mismatch, and repeat the affected scenario after correction.

Related guides