MarTech · Chat Center
A Unified Chat Inbox for LINE, Facebook, and Your Website
The Unified Chat Center puts LINE, Facebook, and website chats in one inbox. Every conversation has a clear owner. AI drafts replies from your organization's knowledge base, and a person reviews and sends. Stalled or risky conversations are pushed to an urgent room. It suits sales and service teams that handle chats on many channels and do not want to lose any customer.
Can I get the specs for this model? And can it ship this week?
Specs drafted from the catalog. A team member needs to confirm the ship date.
Source: Product catalogChecked. I will confirm the date with logistics and send it myself.
Why does a business need a unified chat inbox for LINE and Facebook?
Customers message from many channels: LINE OA, a Facebook page, and your website. Each channel has a different person watching it, so chats are easily lost. Sometimes two people reply at once. Sometimes no one replies at all.
The Unified Chat Center is designed to bring every message into one inbox, give every conversation an owner, and have AI draft replies from your organization's real data. The team can reply faster while people still make every decision.
Unlike typical business chatbots that follow a script, this system draws answers from the Phak AI knowledge base, such as product specs, terms, and questions you have answered before. When a sale closes, the system automatically creates a handoff task for the next step.
What can the Unified Chat Center do?
One message inbox
Chats from LINE OA, Facebook, and the web in one place, with the history of a returning customer visible without switching apps.
Assign an owner to every conversation
Every conversation has one owner with a timer. A customer who already has a salesperson goes to that person's queue first, so two people do not reply at once.
Queue ordered by risk
The queue is not sorted by arrival time. It looks at risk, such as whether it is a public comment, the customer's tone, whether they are new or existing, and how long it has waited.
AI drafts replies from the knowledge base
AI summarizes the conversation and drafts a reply from your organization's data. A person reviews and sends.
Urgent room and alerts
Chats with no reply, long waits, or reputation risk are pushed to the urgent room, and the team lead is notified.
Close the sale, then hand off the job
When a sale closes, the system creates a handoff task for the next step, such as booking a site measurement or issuing a quote.
Executive summary page
Executives see open chats, risky conversations, and jobs already handed off on one page.
When does AI reply on its own, and when does it go to a person?
| AI confidence level | What the system does |
|---|---|
| High: a common question with a clear answer in the knowledge base | Can reply, under the rules your organization turns on |
| Medium | Drafts a reply for a salesperson to review and confirm before sending |
| Low | Always hands off to a person |
| Complaints, price negotiation, or public comments | Always handed to a person, no matter how confident the AI is |
| Final price, discounts, installation dates | The system does not confirm these. A person must confirm them |
The path of one chat message through the system
An example of how a conversation flows, from the customer's first message to closing it out.
Customer messages from any channel
The system sends an automatic acknowledgment so the customer knows the message arrived, and links it to the existing customer profile if there is one.
Match an owner
A customer who already has a salesperson goes to that person first. New conversations are distributed by your team's rules.
Order the queue by risk
Conversations with more potential damage move up first, not the ones that arrived first.
A person replies, with AI drafting
Questions about specs or safety go to a specialist with reference documents. Reports of defective or incomplete deliveries can open a claim directly from the chat.
Close and ask for a rating
After closing, the system asks for a satisfaction score. If the score is low, it alerts the team lead to follow up.
Rules the screen helps your team follow
The system is designed to build key rules into the screen, so they do not depend on each person's memory.
- Staff cannot type a price into a chat. Prices must come from an approved document.
- Questions about structure or safety are answered by a designated specialist with reference documents.
- A claim can only be opened with a real delivery document attached.
- Every conversation has one owner, which prevents duplicate or conflicting replies.
Clear scope
What Phak AI does
- Phak AI helps combine chats from many channels and assign an owner to every conversation.
- Phak AI helps draft replies from your organization's knowledge base for people to review before sending.
- Phak AI helps push stalled or risky conversations to the urgent room and alert the team lead.
- Phak AI helps hand off work to the next step when a sale closes.
What it does not do
- It does not confirm final prices, discounts, or installation dates in place of a person.
- It does not answer complaints or public comments on its own without a person watching.
- It does not promise reply times for your team. Time limits are set by each organization.
- It does not generate images or video.
Frequently asked questions
How is a unified LINE and Facebook inbox different from an auto-reply chatbot?
A typical chatbot answers from a fixed script. The Phak AI Unified Chat Center drafts replies from your organization's knowledge base, assigns owners, orders the queue by risk, and has a person send the important ones.
Will AI reply to all our customers on its own?
No. AI can reply on its own only to high-confidence questions, under rules your organization turns on. Medium-confidence replies need a salesperson to confirm. Complaints, price negotiation, and public comments always go to a person.
What happens if nobody answers a chat?
Every conversation has an owner and a timer. If it stays open past the limit your organization sets, it moves to the urgent room and the team lead is notified.
Where does the AI get the information for its drafts?
From your organization's central knowledge base on Phak AI, such as product data, sales terms, and answers your team has already reviewed. The data stays on a dedicated server for your organization and is not used to train AI models.
How do we get started?
Start by discussing the channels you use and your team's reply rules. The system is then installed on your organization's dedicated server, connected to LINE OA, Facebook, and website chat, and rolled out to your sales and service teams. Talk to the Phak AI team.
Read next
LINE Q&A Assistant
Answers LINE questions from your organization's knowledge. If unsure, it hands off to staff.
Read more →Central Knowledge Base
All your documents in one place. Ask in Thai and get answers that cite the original document.
Read more →Sales Workspace
One place for customer and deal data, with AI to prepare meetings and draft proposals.
Read more →Business Systems
Business Systems for made-to-order work: chat, quotes, contracts, billing, and profit per job in one job record
Read more →Security
A dedicated server per organization, no model training on your data, role-based access, every question logged
Read more →Talk to the Phak AI team
Share your real work, install on a dedicated server, connect existing tools, then add apps one at a time
Read more →Talk to the Phak AI team
Tell us about your business. The Phak AI team will show you the systems that fit and plan the rollout with you, without replacing what you have.