Support App
So you stop answering the same question again and again.
Support is the app for answering your customers’ questions. Questions that arrive at night on LINE, WhatsApp or your website chat are answered on the spot from your company’s knowledge. Conversations from LINE, WhatsApp and email land as cases, and reply drafts follow the way you have answered before. What you resolve goes back into your company’s knowledge as material for the next answer.
What Support does
Receive, answer, keep. That is all it does.
Receive
Conversations from LINE, WhatsApp and email land as cases on their own. No copying into a spreadsheet. Website chats are recorded, and when someone asks to talk to a person, they reach sales as a lead.
Answer
Reply drafts are built from your company’s knowledge and the replies you actually sent before. A person decides whether to send.
Keep
When you record a resolution, the AI drafts what caused the problem and how it was fixed. What you approve becomes company knowledge and material for the next answer.
The problems it solves
How many times this month did you answer the same question?
Measured on a service we actually run, the median time to reply was 22.6 hours, and 72% got a reply within 48 hours. Nobody was slacking — there just weren’t enough hands (397 inquiries that received a reply, as of October 2026).
Features
What you can do with the Support App.
Six features. The AI offers drafts and material, nothing more. Sending and deciding are always done by a person.

Questions on LINE, WhatsApp and your website chat are answered from your company’s knowledge. With the sales add-on, customers who seem ready see your products as cards. Next morning, you just read the list of what was asked.
Screens are drawn for explanation. Details differ from the actual product.
No made-up answers
An AI that can say “I don’t know”.
What makes AI scary in support is a confident, plausible lie. If it says “refunds within 30 days”, you are bound by it. So the facts are fixed by the system and handed over; the AI only chooses the wording.
It becomes an asset
What you answer today helps you next month.
Usually support history just piles up, and nobody rereads it. Here, what you resolve comes back as material for the next answer. The more inquiries you get, the more material you have.
A lesson is drawn out
When you record a resolution, the AI drafts the cause and the fix. No report to write.
Approve it into knowledge
Only what you judge good becomes company knowledge.
Material for the next draft
When a similar question comes in, that knowledge goes into the draft.
Gaps become article ideas
Things asked again and again with no answer on your site become material for proposals to grow your site.
Articles answer at night
Published articles become material for chat and LINE answers. The loop closes.
This loop is connected end to end and actually running. From inquiries imported from past exchanges, 370 lessons have been approved as knowledge, as of October 2026.
What changes
Every answer you give makes the next one easier.
We don’t promise a number of hours saved. What we show here is how it works.
- Night questions do not wait for morning
Questions on your website chat, LINE and WhatsApp are answered on the spot from your company’s knowledge. Next morning, you just look at what was asked.
- Whoever replies, it is one company’s answer
Drafts come from the same knowledge and the same past replies. The owner’s ten lines and a new hire’s two no longer read like two companies.
- No more copying
Conversations become cases on their own, linked to the original thread. The one you forgot to copy is no longer the one you lose.
- More inquiries, more material
In an ordinary help desk, more inquiries only means more cost. Here, the extra goes into your knowledge.
- 22:10A question arrives on LINE: “Is this okay for a first-timer?”
- 22:10An answer comes back at once, from your company’s knowledge. It's built not to force an answer to what it doesn't know.
- 22:10They seem ready to buy, so products appear as cards.
- 22:11They tap through to the product page. The conversation is logged as a case, just as it was.
- Next dayYou read the list of what was asked. The first response is already done.
One question at 10 p.m.
Channels and add-ons
Receive questions where customers already ask.
Connect the places where questions arrive. The conversation channels are website chat, LINE, WhatsApp (an add-on; you bring the Meta account), and your email inbox.
Using it with peace of mind
Four promises that protect your customer conversations.
The AI does not send to customers without your say-so
In your internal screens, the AI only drafts replies. Sending anything to a customer happens only with your approval — or within a scope, budget and deadline you set when delegating to the AI. The exception is answering live conversations on your website chat, LINE and WhatsApp, and there it answers only from what is in your company’s knowledge.
Your history never trains AI
The support history you bring in is used only to find answers inside your company. It never trains AI, and it never mixes with other companies’ data.
Anything that looks personal stops there
When conversations are brought in as material for knowledge, anything that seems to contain personal data goes no further until you have looked. A customer’s private situation never becomes company knowledge without you knowing.
One customer’s story never shows up in another’s answer
Lessons drawn from inquiries are internal-only by default. Your site’s AI assistant is built to answer only from knowledge cleared for publication, so individual customer cases never go public.
FAQ
What people ask.
Not in your internal screens. The AI only drafts; people send and confirm. The only automatic answers at night are live conversations on your website chat, LINE and WhatsApp — and even there, it is built not to answer what is not in your company’s knowledge.
Support is included in the OS iXanagi plan ($299/month per brand, provisional while we tune pricing).
No. For now we recommend running both. Bring in your past history and start with the night-time first response and the loop that turns answers into articles. Move intake over only once you see it working — that way you lose nothing.
We don’t recommend replacing it. Intake, history, categories, priority, deadlines, and drafting and sending replies all work, but it is not the same as the fine detail of a mature dedicated product. The conversation channels are website chat, LINE, WhatsApp and email, and people decide who takes a case (there is no automatic assignment).
No. Your history never trains AI. It is used only to find answers inside your company.
Customer data can be exported and deleted, and that reaches their conversations and form answers too.
It is narrowed so that it does not. Incoming cases themselves never go to the approval inbox. What does is anything read as a possible sales pitch, and candidate lessons. Lessons of the same kind are shown as one bundle; ones whose heading overlaps existing knowledge are shown one at a time. Anything that can’t be decided is not thrown away — it goes to a person.
Honestly, the payoff grows with volume. At a few a month, the night-time first response still helps, but the part where knowledge grows does less. Since you can bring in past history, though, what you already have becomes the first material.
Turn customer support from a cost into an asset.
Every inquiry holds material for the next answer. Start by catching it instead of letting it slip away — beginning with putting your business into words, together with the AI.




