QNE + AI: the second AI you add on top — the one QNE itself will not write
QNE already ships its own AI for general-ledger queries — and it is a good one. Here is what a different layer of AI looks like, and why a KL retail group adds it even after they have the in-product assistant.
minidesk · Editorial team, minidesk
QNE already ships its own AI for general-ledger queries. The in-product assistant is a good one — it reads the QNE ledger, drafts invoice reminders, surfaces cash-flow anomalies, answers the question a finance manager would have to open three reports to answer. If your business is entirely inside QNE, the in-product AI is enough. If your business is QNE plus a POS, plus an e-commerce platform, plus a warehouse system, plus the WhatsApp your customers actually use, the in-product AI is not enough. Here is what the second layer of AI looks like, and why a KL retail group adds it on top of the in-product one.
What QNE's own AI does, and what it does not
QNE's in-product AI is anchored to the QNE interface. You open QNE, you ask the assistant, the assistant reads the QNE data, the answer appears in the side panel. The data the assistant can see is the data inside QNE — the chart of accounts, the invoices, the bank feed, the SST submission. The data the assistant cannot see is the data outside QNE: the WhatsApp thread, the customer's email, the POS, the e-commerce platform, the warehouse system, the project tracker. For a business that runs its world inside QNE, that is fine. For a business that runs its world across QNE and four other tools, it is not.
The second layer of AI is the layer that reads the data outside QNE and writes the answers into the channel the customer is actually using. The second layer does not replace the first — your bookkeeper keeps the QNE workflow they already know — it adds the channel, the language, and the cross-software view that the in-product assistant was never built to cover. The two AIs do different jobs. The business gets both.
Three concrete use cases
1 · The multi-entity retail group in KL
Farah is the finance manager at a 6-shop retail group in KL. The group runs QNE for the consolidated ledger, a separate POS for each shop, a third-party e-commerce platform for the online channel, and a warehouse in Subang for fulfilment. The four systems do not talk to each other. The question Farah gets most often from the group CEO is the cross-system one: "what were total sales yesterday, across all channels." The answer today is three logins, two CSV exports, and a VLOOKUP. The CEO has stopped asking because the answer takes 20 minutes.
With the second layer of AI, the cross-system question is one ask. The CEO opens the WhatsApp chat: "What were total sales yesterday, across all channels?" The AI reads the QNE consolidated ledger, the POS daily close, the e-commerce platform orders, and the warehouse fulfilment feed, and replies: "Yesterday: RM 84,200 total — RM 52,400 in-store POS (top shop: Bangsar, RM 18,200), RM 28,300 e-commerce, RM 3,500 corporate catering. The biggest single order was a RM 4,200 e-commerce order from a Petaling Jaya customer. The 4 pending fulfilment orders are worth RM 1,820." The CEO forwards the answer to the group chat. The 20-minute VLOOKUP is gone.
2 · The F&B chain owner who wants WhatsApp, not the QNE UI
Daniel owns a chain of 4 F&B outlets in the Klang Valley. The chain runs QNE for the books, a separate POS for each outlet, and a delivery platform for the online orders. Daniel's customers are on WhatsApp. Daniel's staff are on QNE. The two never meet. Today, every customer WhatsApp is answered by a human, who has to log into QNE, look up the order, copy the number, paste it into WhatsApp, and reply. The cycle takes 3 minutes per message. The chain's two customer-service staff spend the entire day on the cycle.
With the second layer of AI, the WhatsApp is the assistant's inbox. A customer in Petaling Jaya sends a WhatsApp at 7:42pm: "Where is my order #DD-2025-0812?" The AI reads the QNE sales order, the POS order log, and the delivery platform status, and replies: "Hi Encik Lim, your order DD-2025-0812 (RM 48.50, 2x nasi lemak, 1x teh tarik) was dispatched at 7:15pm and is 8 minutes away. Reply here if you need to call the rider." The reply lands at 7:42:18pm. The customer reads it on the train home. The two customer-service staff are no longer the bottleneck.
3 · The IT distributor who wants the cross-software view
Hafiz runs a small IT distribution business in KL. The business runs QNE for the books, a separate inventory system for the warehouse, and a third-party e-commerce platform for the online channel. The question Hafiz gets most often from his 3-person finance team is the cross-software one: "what sold yesterday, what was shipped, what is still in the warehouse." The answer today is three logins, two CSV exports, and a VLOOKUP. The team has stopped asking because the answer takes 20 minutes.
With the second layer of AI, the cross-software question is one ask. Hafiz asks: "What sold yesterday, what was shipped, what is still in the warehouse?" The AI reads the QNE sales log, the inventory system, and the e-commerce platform, and replies: "Yesterday: 47 orders for RM 38,200. 41 were shipped (QNE invoice raised, warehouse confirmed). 6 are still in the warehouse — 4 in KL, 2 in PJ. The 6 orders are worth RM 4,820. The oldest unshipped order is 2 days." Hafiz forwards the answer to his team. The 20-minute VLOOKUP is gone.
Why QNE will not (and should not) ship the second layer
QNE's in-product AI is a good assistant for the work that lives inside QNE. The second layer of AI is a different category — the cross-software view, the WhatsApp channel, the language-aware reply in Bahasa Malaysia, English, Mandarin, Tamil. QNE's product roadmap is the cloud accounting suite, not the cross-software AI for every tool a Malaysian SME runs. They will not ship it, because the second layer is a different product category from the ledger, and QNE's job is to be the ledger.
That is not a criticism. It is a description. The QNE customer is best served by a layer that lives on top of QNE, not by a vendor whose roadmap is the cloud suite. The second layer of AI is exactly that — it reads the QNE ledger QNE is already serving, and it adds the channel, the language, and the cross-software view QNE will not ship. The two AIs answer different questions. The business gets both.
“I was already paying for the QNE in-product AI. What I was missing was the AI that talks to my customers in WhatsApp — and the AI that knows what the POS number is when the CEO asks about yesterday's sales.”
— Farah, finance manager at a 6-shop retail group in KL
What a normal Tuesday looks like with both AIs on
It is 7:30am in the PJ office. Farah is at her desk, the WhatsApp is pinging with overnight messages from the 6 shop managers. The QNE in-product AI has already categorised the bank feed overnight and surfaced a cash-flow anomaly: the Subang shop's deposit was RM 4,200 short of the expected amount. The second layer of AI has already read the WhatsApp inbox, answered the 4 questions that were answerable from the QNE data alone, and queued 3 follow-ups for Farah's review. Farah opens the queue at 7:35am, approves the 3 messages, drills into the Subang anomaly in QNE, calls the shop manager. The cash-flow anomaly is a till miscount — recoverable, will be reconciled by 10am. The morning is done by 8:15am, the actual finance work starts at 8:30am, the CEO has the daily summary in his WhatsApp at 8:45am.
What it costs, in real money and in real time
The second layer of AI is a flat monthly fee — the number is on the pricing page, the quote is the quote, no per-message markup, no per-prompt surcharge, no enterprise tier. The OAuth for QNE is included. The cross-software connectors for the POS, the e-commerce platform, and the warehouse are included. The WhatsApp Business API access is included. The Meta conversation fees are the same fees every platform pays, and we do not add a per-message markup on top. The only other line item is the AI usage, which is metered in the same units every modern AI tool uses, and which your finance team will not notice unless they are doing something unusual.
The time saving is harder to put a number on, because the time is not a single block — it is a thousand two-minute savings across a month. The Monday morning cross-software report that used to take 20 minutes of VLOOKUP. The Friday afternoon invoice chase that used to take 90 minutes. The "any update on my order DD-2025-0812" reply that used to take 3 minutes per message. The "what sold yesterday" question that used to take 20 minutes. The second layer of AI takes each one of those from minutes to seconds, and the sum is the difference between a finance team that is drowning in WhatsApp and a finance team that is using WhatsApp to grow.
The security question every Malaysian finance manager asks first
The first question every Malaysian finance manager asks, after "what does it cost" and "does it work with my software", is "where does my data go." The honest answer: the QNE ledger stays in QNE. The second layer of AI requests only the OAuth scopes it needs (read invoices, read contacts, read reports) and never writes back. Only the answer leaves the system, not the row, not the table, not the credentials. The AI sees only what your QNE user is allowed to see, enforced at the OAuth grant, not in a prompt. The data isolation is the same isolation QNE itself enforces — your bookkeeper cannot see an entity she does not have permission for, and the second layer of AI cannot see an entity your bookkeeper cannot see.
How to add the second layer of AI to your QNE
The install takes about twenty minutes. The OAuth is one click in the QNE consent screen. The cross-software connectors are added one at a time, in any order. The WhatsApp is one Meta Embedded Signup window. The AI is the same AI that runs every other connector on minidesk — once QNE is connected, the AI knows how to read it, and the second layer of answers is in the chat.
- 1Sign up at minidesk.co and pick QNE as your connector.
- 2Click the OAuth link, sign into your QNE account, and approve the minimum scopes the second-layer AI needs (read invoices, read contacts, read reports). Your in-product QNE AI keeps doing what it was already doing.
- 3Add the cross-software connectors — POS, e-commerce, warehouse, whatever your business runs alongside QNE. The AI reads each one in turn, and the cross-software answers start working the moment the second connector is live.
- 4Connect your business WhatsApp number through Meta Embedded Signup — minidesk holds the Meta Business verification, so the typical 2–4 week wait is gone.
- 5Ask the AI a question in plain language. The first answer takes a minute because the AI is mapping your chart of accounts. After that, every answer is under three seconds.
- 6Hand the AI a workflow ("every Friday at 4pm, draft a WhatsApp for every customer whose invoice is over 30 days past due") and it does it, shows you the queue, and asks before anything goes out.
If your business is already running on QNE, the second layer of AI is the smallest change you can make to the largest difference in how the day feels. QNE stays QNE. Your bookkeeper keeps the in-product workflow she already knows. The cross-system question gets an answer in 4 seconds, the WhatsApp reply lands in the customer's own language, and the Monday morning report is in the CEO's hand before his coffee is cold.
→ See the QNE integration: /integrations/qne