01
What is this chatbot?
An analytics assistant that answers natural-language questions about your POS (point-of-sale) data.
Your stores' sales, transaction counts, average spend, best-selling products, and hourly patterns are already in the system. Until now you had to pick a period and a store on the dashboard yourself — now you can simply ask, “How were the Suyeong store's sales last month?”
✅
The numbers always match the dashboard. The chatbot pulls its figures from the same aggregation engine the dashboard uses. You will never see different sales numbers on different screens.
🔒
Only stores you have access to are included. Answers cover only the stores assigned to your login account. Even if you ask to “show other stores too”, anything outside your permissions is quietly excluded.
02
What can you ask?
The main analysis areas the chatbot covers.
📈
Sales · Transactions · Avg. Spend
Total sales by period, transaction counts, average spend, growth rates, daily trends, and day-of-week patterns.
🏪
Store Comparison · Rankings
Compare multiple stores at once and rank them by sales, average spend, or growth rate.
🛒
Product · Category Rankings
See best-selling products, top categories by revenue, and each item's share of sales.
⏰
Hourly Patterns
Find out when sales concentrate during the day, in 2-hour time slots.
👥
Member Age · Gender
Analyze which categories members buy most, by age group and gender.
📝
Report Interpretation (the “why”)
Beyond the numbers — for “why did this happen”, trend interpretations and notable events are found in existing reports and answered with sources.
03
Ask it like this
You don't need perfect sentences. Ask the way you would normally speak.
Tell me the Suyeong store's sales and average spend last month
Rank all stores by August sales
Top 10 best-selling products at the Gwangan store over the last 3 months
What time of day is the Suyeong store busiest?
Which categories did female members in their 20s buy most last week?
Find in the reports why sales dropped last month
You can keep the conversation going
The chatbot remembers the conversation so far. You don't have to repeat the store or the period every time.
Show me the Suyeong store's August sales
What about the next month? → continues with Suyeong, September
And the Gwangan store? → keeps the period (September) and switches the store
Compare the two → compares Suyeong and Gwangan for September
04
Getting the best answers
Keep a few things in mind and you'll get accurate answers much faster.
- Mention the store and the period together — “Suyeong store / last month”
- Relative expressions like “last month”, “the last 3 months”, or “the first half of this year” work as-is
- When comparing stores, name the targets — “all stores”, “store A and store B”
- Narrow down with follow-up questions — “Which one ranked first?”
- Curious about the ‘why’? Ask it to “find the reason in the reports”
⚠️ These are hard to answer
- Single-receipt questions — “How many receipts over ₩30,000?”
- Payment-method breakdowns — “What's the card vs. cash ratio?”
- Questions about the future or periods outside the available data
- A first question with no store and no context (“Show me sales”)
- Sales forecasting or anomaly detection
💡
If the store or period is missing, it asks back. When the chatbot can figure things out on its own, it just answers; when it's ambiguous, it briefly checks — “Which store should I look at?” A one-word reply is fine.
05
Understanding the answer screen
What the elements that appear alongside answers mean.
Progress
“Fetching sales…” — while the answer is being prepared, you can see in real time which data is being retrieved. The finished answer follows shortly.
Tables · Charts
Numbers come as tables and charts — sales trends and store comparisons are shown as tables and charts, not just text.
Shortcuts
Jump-to-detail buttons — answers about sales, products, and store comparisons include buttons that take you straight to the related analysis screen. Use them when you want to dig deeper.
Sources
Report citation chips — ‘why/interpretation’ answers show the daily, weekly, or monthly report they are based on.
06
Limitations to keep in mind
For accuracy's sake, here is what the chatbot currently does not cover.
🧭
If a number looks off, just ask. You can ask “How was this number calculated?” or “Which stores does this cover?” — the chatbot states which stores and periods its answer is based on.
07
Three languages supported
한국어 · English · 日本語 — chatbot answers and AI reports are available in three languages.
🇺🇸
English
“Show me last month's sales for the Suyeong store.”
Answer language
Follows your ‘AI response language’ setting — the language of chatbot answers and AI reports is the one chosen at Settings → AI Language → AI response language. Once set, it is saved to your account and kept.
Display language
Separate from the display language — the language picker at the top of the screen (globe icon) only affects on-screen text such as menus and buttons. For example, you can browse the screen in 日本語 while receiving answers in English.
💡
You'll be notified when the two differ. If you enter the chat screen while your display language and AI response language don't match, it asks once whether to switch the AI response language to the current display language. Pick whichever you prefer.
08
Daily · Weekly · Monthly reports
Based on the same data as the chatbot, AI analysis reports are provided for each store in three cadences.
📅
Daily Report
Summarizes each day's sales and transactions, highlighting notable changes and events.
🗓️
Weekly Report
Reviews the week's flow with day-of-week patterns and interprets week-level trends.
📊
Monthly Report
Wraps up the month with a sales summary, trend interpretation, and notable events.
Where
The ‘AI Analysis Report’ menu in the sidebar — pick a store and a report type (daily, weekly, monthly) to see the latest report right away.
Past reports
Browse the calendar — in the ‘Past Reports’ tab, pick a date on the calendar to reopen past daily, weekly, and monthly reports.
Email
Get it in your inbox — reports can be sent to your account email, so you can read them without opening the dashboard.
Languages
Reports come in three languages too — report content also follows your ‘AI response language’ setting: 한국어 · English · 日本語.
🔗
The chatbot and the reports are connected. Ask the chatbot something like “find in the reports why sales dropped last month”, and it searches exactly these daily, weekly, and monthly reports for evidence, answering with the source.
09
Frequently asked questions
Can I trust the numbers in the answers?
Yes. The chatbot pulls its figures from the same aggregation functions as the dashboard. For the same store and period, values are designed and verified to always match the dashboard.
Do I need to know the exact store name?
Aliases work. For example, “L2 Market Suyeong” is also recognized as “Suyeong store” or simply “Suyeong”. If the store is ambiguous, the chatbot shows candidates and asks.
Do expressions like “last month” or “the last 3 months” work?
Yes. They are converted to exact date ranges based on today's date. You can also specify an explicit range (e.g., March 3 – April 17).
Does it remember earlier messages?
Within a single conversation it keeps the recent context (store, period), so follow-ups like “what about the next month?” or “and the Gwangan store?” flow naturally. Start a new conversation when moving to a new topic.
Does it also explain the “why”?
Yes. Beyond retrieving numbers, interpretations such as “why did sales drop” are found in the existing daily, weekly, and monthly reports and answered with sources. The numbers themselves are always computed exactly from the aggregated data.
Can I use it in Korean or Japanese?
Yes. Three languages are supported: 한국어 · English · 日本語. The language of chatbot answers and AI reports can be changed at Settings → AI Language → AI response language, applied independently of the display language.
Can I see other people's store data?
No. Only the stores assigned to your login account can be queried. Stores outside your permissions are never included in answers, however the request is phrased.