What this brand uses
The brand first. According to the homepage of REX Instrument's own website, it supplies non-destructive testing instruments: industrial borescopes, infrared thermal cameras, concrete moisture meters, leak detectors, and instruments for leak tracing and home inspection. The site also shows an instrument rental service. The use cases it lists run from leak tracing, home inspection, exterior wall repair and plumbing to car and motorcycle repair, refrigeration and air conditioning, and semiconductors.
Put differently, the people who write in are mostly professional buyers with a job to do. For specifications and prices, go to its website; this article does not restate them.
This table is the only "case fact" in the article.
| Channel | Product | Since | Type of work handled |
|---|---|---|---|
| Facebook Page messages | FRIDAI Chat | 2026 | DM auto-reply, knowledge-base answers, human handoff |
| Instagram messages | FRIDAI Chat | 2026 | DM auto-reply, knowledge-base answers, human handoff |
| Threads | FRIDAI Social | 2026 | Comment replies and social patrol |
The last column describes what the two products are built to do. It does not say which features REX Instrument has switched on or what rules it has written; that is the brand's own business and we do not cover it.
For how other brands combine channels, see the FRIDAI customer stories roundup.
Why a business like this would set it up this way
This section is FRIDAI's own view, reasoned from the business type (selling industrial products to professional buyers). It is not a reason REX Instrument gave us.
B2B messages open with specifications and quotes
A cake shop gets "is it in stock, how much, when can I collect". A brand selling industrial equipment gets something more like: how narrow is the pipe, how long, is the environment wet or hot, buy or rent first, what kind of invoice. Each question affects product selection, and one wrong condition invalidates the whole quote.
So in our view, the first goal of DM auto-reply for industrial products is not "answering" but "asking the right questions". The AI collects the use case, environment, quantity and contact details and passes them to someone who knows the product. Sales then receives an inquiry they can answer directly, not a bare "how much is it".
The AI should not be the one to quote
B2B prices usually depend on quantity, accessories, lead time and whether it is a purchase or a rental, and may vary by customer. Letting the AI state a number carries two risks: if it is wrong you have to clean up, and even if it is right it may have skipped a condition that should have been asked. FRIDAI itself notes that AI output can be incomplete or inaccurate and that customer commitments need human review.
The steadier split: the AI can point to information that is public on your website; any unpublished price, discount or lead time goes to a person.
Threads is where you are seen; DMs are where business is discussed
Many manufacturers and B2B brands assume social media is not for them. But professional buyers scroll Threads too, and problems on site and tool experiences are exactly the kind of thing people discuss. A brand has a presence only when its public comments get a response; real inquiries are then steered to DMs or the contact channels on the website. Comments and DMs are two different jobs, which is why two products are involved: comments belong to FRIDAI Social, DMs to FRIDAI Chat.
How a similar brand could set this up
The following is general advice for B2B, manufacturing and industrial product brands. It is not REX Instrument's internal setup and does not represent any result.
| What to prepare | Why | Guide |
|---|---|---|
| Answers to nine questions before you start | Confirm volume, who takes over, and what the AI must not answer | AI customer service implementation: 9 questions |
| An inquiry requirements list | The AI asks from the list so sales gets a complete inquiry | What is conversational commerce |
| A fixed handover line | Customers learn who will reply and roughly when | Customer service reply templates |
| A split between comments, DMs and posting | Three jobs with different rules; do not merge them | What is social media automation |
| Instagram auto-reply tools and limits | Know what Meta's built-in features do before adding a tool | Instagram auto-reply tools compared |
Step 1: break an inquiry into fixed fields
Ask sales to list "what we must know before answering an inquiry". It is rarely more than five items. That list is the AI's task and also its boundary: collect, hand over, say no more.
Below is a requirement-collecting message you can adapt; swap the fields for your industry's terms. The template is in Chinese because that is what your customers will read; it thanks the customer and asks for the object and environment to be inspected, the conditions that matter (such as size range, length or precision), quantity and buy-or-rent preference, and company name plus contact details, then says a specialist will reply with a quote and advice.
您好,感謝您的詢問。為了讓負責的同事給您正確的建議,想先請教幾件事:
1. 您要檢測或處理的對象與使用環境
2. 目前在意的條件(例如尺寸範圍、長度或精度)
3. 預計的數量,以及偏好購買或租賃
4. 公司名稱與方便聯絡的方式
收到後會由專人回覆您報價與建議,謝謝。Step 2: write down what the AI must not say
For industrial products, "must not say" matters more than "may say". Include at least these: do not quote prices that are not public on the website; do not promise lead times or stock; do not judge that a model will definitely suit the customer's site; do not answer operating questions that involve safety or regulations. On these topics the AI's job is to say a specialist will take over, then hand over.
Step 3: give public information to the knowledge base
Opening hours, locations, service area, whether you rent equipment, how to send something in for repair, product introductions already public on your website: repetitive questions that are hard to get wrong belong in the knowledge base for the AI to answer. Sales time should go to selection and quotes, not to "where are you located".
Step 4: use a different set of rules for comments
Public comments are visible to everyone, so replies should be more conservative. The general advice: respond to technical discussion and add public information, steer price inquiries to DMs, and alert a person for negative or disputed comments. Early on, have a person review replies before they are sent, and relax that once you trust the setup.
What to leave to people
In our view, B2B and industrial product brands have five kinds of message the AI should not close on its own: quotes and discounts, lead-time and stock commitments, product selection and suitability, repair and warranty disputes, and complaints. Every paid FRIDAI plan includes human handoff and human review, and can alert the team through FRIDAI's official LINE account when someone needs to step in.
This chart is a generally recommended division of work, not REX Instrument's actual flow. There is one point to it: every path ends with a person, and the AI's job is to make sure that person has complete information when they take over.
What this article does not cover
There are no reply rates, inquiry counts, sales or revenue figures here, and no customer quotes. The reason is simple: those are the brand's own operating information, and B2B sales cycles are long while inquiry volume and order size vary widely, so someone else's number would not hold for you. We would rather leave it out than hand you a good-looking figure you cannot verify.
Also not covered: REX Instrument's plan tier, fees, rule content or conversations, or its product specifications and prices.
So how do you judge whether this fits you? Test with your own inquiries. FRIDAI offers a trial for a US$1 verification fee (NT$30 with PAYUNi or JKO Pay, two Taiwan payment methods): 30 days with 100 AI replies. It converts to the plan you chose when the 30 days end or the trial allowance runs out, and it is limited to one trial per service per email address. B2B message volume is usually modest, so 100 replies are enough to see the direction. Record three things: how many inquiries arrived with requirements fully collected, how many handovers went to the right person, and whether the AI ever stated a price or made a commitment it should not have. If the third happens even once, go back and fix the rules first.
Two more cases in this series cover consumer brands, useful as a contrast with B2B: the gogoout car rental platform case and the 4K Baking case. For multiple channels and multiple front doors, see the ON LIVE case.
FAQ
Which channels does REX Instrument use FRIDAI on?
In 2026 REX Instrument (雷克斯
Why are there no results figures in this article?
Results are the brand's own operating information and FRIDAI has not published them. B2B inquiry volume, order size and sales cycle vary widely, so another brand's numbers do not transfer. The most reliable check is a 30-day trial with your own inquiries.
We get few B2B messages. Do we still need AI customer service?
It depends on what a missed inquiry costs you. Brands with few but high-value messages usually care less about saving labour and more about having evening and weekend inquiries caught and the details collected. If you already reply promptly, you can wait.
Can AI customer service give a quote directly?
Technically it can reply with public prices written in the knowledge base, but for B2B we do not recommend letting it state unpublished prices, discounts or lead times. The safer approach is a rule that sends these to a person while the AI collects requirements and says a specialist will reply.
Is Threads worth it for a manufacturer?
You have to verify that yourself; we have no guaranteed answer. What is certain is that unanswered public comments amount to no presence at all. If you go ahead, decide which comments get a public reply, which go to DMs and which alert a person, then consider a tool to share the load.
Are comment replies and DM replies the same product?
No. FRIDAI Chat handles DMs; FRIDAI Social handles comment replies, comment-to-DM flows and social patrol. According to FRIDAI's public notes, comment-to-DM needs a FRIDAI Chat plan on the same account, and Threads DMs are a separate yearly plan.
How FRIDAI can help
If your DMs also open with specifications and quotes, start with an inquiry requirements list: let the AI collect details and point to public information, and leave quoting and selection to sales. On the comments side, use a separate, more conservative set of rules.
- Product: FRIDAI Chat
- FRIDAI Social: FRIDAI Social
- Customer stories: FRIDAI customer stories from Taiwan brands
- Before you start: AI customer service implementation: 9 questions to ask first
- Instagram: FRIDAI on Instagram
- Threads: FRIDAI on Threads
- Facebook: FRIDAI on Facebook
Official sources
- REX Instrument (雷克斯
儀器) homepage: rexinstrument.com.tw - FRIDAI website (customer logo wall): fridai.fansnetwork.ai/
- FRIDAI Chat: fridai.fansnetwork.ai/
fridai-chat - FRIDAI Social: fridai.fansnetwork.ai/
fridai-social - FRIDAI product and pricing summary: fridai.fansnetwork.ai/
llms.txt