Customers in Singapore often message a business on WhatsApp, and the messages keep coming after the shop, clinic or centre has closed. A parent asks about a trial class at 10pm. A patient asks for a slot at 9pm. By the time someone replies the next morning, the customer has often booked somewhere else.
An AI chatbot on WhatsApp can answer those messages at night. The useful question is what it answers from, and who checks it before it speaks for your business.
What Glimmer does in your WhatsApp
Glimmer is the AI agent Glints builds for Singapore companies, and the one we run our own company on. For customer messages, it reads your business WhatsApp line and works through the queue.
- It answers from your own documents. Opening hours, prices, the return policy, the class schedule. If the answer is not in them, it does not guess.
- It checks your systems before it replies. It can look up live stock counts, an order's status or open slots in your calendar, so the reply is about this customer's order, not a general one.
- It replies in the customer's language. A message in Mandarin or Malay gets a reply in the same language.
- It holds what it cannot check. A discount it cannot find, or a question only the doctor can answer, gets a short holding reply and goes to a person.
In the morning, you ask Glimmer what came in overnight. It lists what it answered, what it held for you and why.
You choose what needs a person's OK
A WhatsApp reply goes out under your company's name, so you decide what Glimmer may send on its own. You set a rule once for each kind of message: "Ask first" or "Just do it".
For example, a clinic might let Glimmer answer questions about opening hours on its own. Booking requests get a slot held in the calendar, and the front desk confirms each one in the morning. A catering quote or a refund stays a draft until the owner taps send.
When Glimmer needs an OK, it asks your team in your team chat. Nothing in the "Ask first" group goes out until someone approves it. See how approvals work.
Example chats by industry
Each of these is an example chat on our solutions pages, with the job it takes off the team:
- A clinic's 9pm booking requests, with slots held for the front desk to confirm.
- A tuition centre's late-night parent enquiries, answered from the term schedule.
- An online shop's weekend messages, across WhatsApp, Instagram and email, cleared before Monday.
- A catering enquiry at 11pm, drafted for the owner to send that night.
- Stock questions from four inboxes, answered from live stock counts.
What it does not do
Glimmer reads only the WhatsApp line you connect to it. Chats on a staff member's personal phone stay outside it.
It also does not replace the person who knows your customers. The routine questions get answered at night. The unusual ones reach a person in the morning, with the whole thread attached.
How this differs from a basic chatbot
A basic chatbot answers from a fixed list of questions and replies. Glimmer works inside the systems you already run, such as your calendar, your POS, Shopify or Xero. Your team talks to the same Glimmer in Lark, Microsoft Teams, Slack, Google Chat, WhatsApp or Telegram.
It runs in a cloud account in your company's name, and everything we build belongs to you.
Grants
Customer service is one of the common uses that IMDA's GenAI x Digital Leaders grant looks for. Other routes may fit too. Check which grant fits your project on our grants page, and apply before you sign anything.
Where to start
For one week, count the WhatsApp messages that arrive after closing time, and how many of them wait until morning. Then book a call and bring that count. We will tell you what Glimmer could answer on its own and what should wait for you.