Konkui AI

Thai online sellers get customer questions on LINE, Facebook and Instagram at once, and most of them repeat: price, stock, shipping and store policy, answered by an admin one message at a time. Handing that to an AI is risky, because a wrong price or stock level does more damage than no answer at all. The brief: an assistant that answers from the shop’s own data from day one, never states a number it didn’t look up, and hands the chat to a person the moment it can’t help.

The hard part of a sales chatbot isn’t answering, it’s not answering wrong. So the data the shop enters is the only source of truth, a second AI reviews every reply before it goes out, and we designed the product around one customer message: arriving in the Inbox, getting an answer, reaching a person when needed, and becoming Knowledge for next time.

1. 1. Set up: each shop in its own Workspace

   Each shop gets its own Workspace with data fully separated from every other shop. The team is made up of Admins and Operators, who sign in with Google.

2. 2. Connect channels: LINE, Facebook and Instagram, by each platform’s rules

   A LINE OA connects with the shop’s account details and only shows as Connected once a test message from LINE is verified; Facebook Pages and Instagram connect through Facebook login. On LINE the Assistant answers within the free reply window before falling back to a separately sent message, and on Facebook and Instagram it only sends inside Meta’s 24-hour window.

3. 3. Teach it the shop: data it can trust

   A shop keeps Products, Promotions, Bookings and Customers. Key fields such as price and stock are labelled so the Assistant knows what they mean, and Promotions apply automatically within their dates. The Assistant can save Bookings and Customers but never search them, so one customer can never be shown another’s details. Knowledge can come from seven kinds of source, from documents and web pages to a Q&A sheet, and any text that looks like a hidden instruction to the AI is flagged.

4. 4. Answer: a second AI checks first

   The Assistant finds the shop’s Knowledge by meaning, in Thai and English. Before it saves anything and before a reply goes out, a second AI checks it: a reply quoting a price or stock level that didn’t come from the shop’s data is sent back for one fix, and if the check can’t run, the chat goes to the team.

5. 5. People step in, and corrections become Knowledge

   Once a person holds the chat, the Assistant stays silent until it is handed back. A thumbs-down with a correction goes into a queue an Admin approves into a Q&A pair. The Curator turns context an Admin pastes into proposals to review, and other AI tools the team uses can propose Knowledge too, but every proposal waits for a person. The Playground shows, for each reply, what it drew on, what the checker decided and what it cost.

Konkui AI runs in production. LINE OA connects today; Facebook Pages and Instagram accounts belonging to other businesses show ‘Review pending’ until Meta’s App Review is complete.

• Every chat in one Inbox

  The team sees every conversation and where it stands, whether with the Assistant, with an Operator, waiting or resolved, and each shop’s data stays in its own Workspace.

• Answers grounded in the shop’s data

  The Assistant answers product, price and promotion questions from what the shop entered, and says so plainly when a fact is missing instead of guessing.

• A person takes over cleanly

  On handoff the Assistant stops replying, but drafts a suggested reply in the composer that the Operator can use, edit or discard.

• Corrections become Knowledge

  Once approved, a team correction is used by the Assistant from the very next message.