Customers get answers from information you have approved; hard questions go to a person.
In development for a restaurant. Not yet live for real customers.
Demo video in preparation
Meanwhile, see the steps and limits below.
The problem today
Customers message through the website, Zalo and phone. Evening messages wait until morning; questions about prices or availability get no firm answer.
How it works
- Step 1Receive questions on the website.
- Step 2The assistant answers from content you have approved: menu, opening hours, policies.
- Step 3Anything outside the approved content, or a complaint, goes to staff with the last few messages.
- Step 4Table booking: the assistant checks availability and creates a booking request for staff to confirm.
Which systems can be connected?
- Website (chat on the page)In development
- Zalo OANot built yet
- WhatsAppNot built yet
- Facebook MessengerNot built yet
Verified: already running with that system. In trial: running on a test version. In development: code exists, not running yet. Not built yet: no code yet. Needs a survey: depends on your system, we must read it before promising. Not supported: not possible yet.
What the machine does, what people do
Machine
Answers from approved content, checks availability, creates booking requests, hands over when out of scope.
People
Approves the content the assistant may use, confirms bookings, handles exceptions.
Rule: the assistant only uses prices, availability and figures that are in the data; otherwise it hands over to a person.
Trial scope
Start with the website and a list of frequent questions. Measure: how many questions get answered and how many go to a person. Timing and cost: quoted after the survey.
Results and evidence
No results yet: not live with customers, nothing measured.