Translated from the Korean original. Some screenshots show the original Korean interface.
In an incident, the longest part is finding the cause
Monitoring screens already collect plenty of what you need: CPU and memory, sessions and SQL, alerts and response times. Yet when an incident hits, operators go through the same routine every time.
“Where do I start?”
“Is this alert the cause or a symptom?”
“Which screen showed that slow SQL again?”
The problem is not a lack of data. It is that connecting information scattered across many screens and narrowing down the cause still depends on human experience.
A seasoned operator finds the right screens quickly and narrows down the cause within minutes. A less experienced one can spend a good amount of time just deciding where to look first. At night or on weekends, when experienced staff are not around, the initial response can be even slower. And once the incident is under control, there is still more work: summarizing what happened, explaining the cause, and documenting the actions taken.
exemONE AI Assistant turns this operational work into a single conversation that finds, analyzes, and summarizes.
1. Four ways exemONE AI Assistant changes operations
1-1. Ask a question and see current operational data right away
Operations teams no longer need to remember menus or memorize metric names. Ask in everyday language, like the examples below, and the AI queries current operational data and shows what you need right away.
- “Show me the recent alerts.”
- “Which service has the slowest response time?”
- “How is the system doing right now?”
The AI queries the operational data currently being collected and presents results that match the question as tables and summary cards. Recent alerts come with their time, severity, and target system. If several alerts relate to the same incident, it groups them into a single issue so they are easier to understand.
Every number in an answer comes with the time it was queried, and clicking a target name takes you straight to the related screen. You can ask about many things: overall system status, slow queries, application response times, transactions, and more. Even if you don't know the exact metric name, just ask in your own words and the AI finds and connects the right data.
1-2. It goes beyond the answer and opens the analysis screen
Most AI chatbots stop once they answer a question. exemONE AI Assistant carries on into the analysis work that follows.
For example, type “Analyze slow queries” and the AI moves to the relevant analysis screen, selects the target and time range automatically, and opens the screen with the query already run. The operations team sees the results the moment they land on the screen.
It supports the main analysis screens, including slow queries, session lookup, performance trends, execution plans, and application transaction trends. From a target list screen, it can also jump straight to the detail tab of the instance you want.
You can also continue the conversation on the screen you are looking at. Type “Analyze this screen” and it reads the charts, lists, and filter conditions on the current screen as they are, explains them, and understands follow-up questions in the same context.
💡 In short, you don't have to remember the menu structure. A conversation alone takes you to the analysis step you need.
1-3. It narrows down to the cause, not just the symptom
Confirming that a service is slow does not solve the problem. What matters is quickly finding which system has the problem and what caused it. exemONE AI Assistant analyzes the cause in the following order.
① Check for anomalies
It compares the target's alerts and key performance metrics with its normal operating patterns to decide whether something is off. It looks not only at averages but also at sudden spikes, so short bursts are not missed.
② Verify candidate causes
Based on the anomaly signals, it checks possible causes one by one. It separates confirmed facts from unlikely items, so you can decide based on data rather than guesswork. It then narrows the cause down to the level where action can be taken, such as the session holding a lock or the SQL causing a slowdown.
③ Analyze the scope of impact
When needed, it analyzes the application and the database together to find which layer the problem started in. For example, when service response is delayed, it shows whether the real cause is a database lock or slow application processing. Instead of the operations and development teams each guessing at the cause, they can respond quickly from the same data.
1-4. Repetitive product tasks, done through conversation
.jpg?table=block&id=210177f0-99c6-47ca-938b-66799c16cec6&cache=v2&width=1200)
Managing the operating environment takes as much time as incident response. exemONE AI Assistant lets you handle these repetitive tasks through conversation alone.
✅ Create dashboards - “Create a MySQL monitoring dashboard”
No need to find and add widgets one by one. The AI picks the performance metrics you need and builds a dashboard that is ready to use.
✅ Recommend alert thresholds - “Recommend alert thresholds for this instance”
It suggests warning and critical thresholds based on the target's actual statistics. One click registers the suggested values in the alert settings screen.
✅ Write operations reports - “Summarize the current status as a report”
The AI pulls together the current operating status into one report with key metrics, root cause analysis, supporting data, and recommended actions. You can save it as a PDF or use it directly as reporting material.
2. How we make AI answers trustworthy
A wrong answer can lead to a wrong action. That is why exemONE AI Assistant is designed so you can check the basis of every answer, as shown below.
Sources shown | Numbers in an answer show which data was queried and when. Reports list the basis for each judgment as a separate item. |
Consistent judgments | Severity is judged by fixed criteria. The same situation gets the same conclusion, and the answer does not change each time you ask. |
Permissions unchanged | It only queries what the user is allowed to see, and targets outside their permissions are left out of answers. |
Honest when it doesn't know | When data is missing or collection was interrupted, it says so instead of concluding "no issues." |
Whether to use the feature and how far, including activation, which AI model to use, and history retention policy, can be controlled by administrators to fit their environment.
3. Use cases
- When you get a call that the service is slow
Type “Summarize the payment service status as a report,” and shortly after you get a report showing which metrics crossed their thresholds, what the cause is, and what action to take. You can act first without wandering through screens.
- When team members disagree
Say the app owner and the DB owner disagree. Ask the AI to analyze both targets together, and it shows with data which side is the cause and which is the symptom. You can decide who should respond based on evidence, not guesswork.
- When a new team member is on call
Even someone still getting up to speed who doesn't know the menu structure can type “Analyze slow queries” and get the analysis screen opened with the query already run.
- When you add a new system to monitoring
Type “Create a monitoring dashboard” to set up the screens, then “Recommend alert thresholds” to get thresholds that fit the operating history and register them. This cuts the time initial setup used to take.
4. An assistant always open next to your monitoring screen
.jpg?table=block&id=2cc53478-05bb-4dcb-a45c-36aa8cbb5f98&cache=v2&width=1200)
Operations teams should spend their time analyzing incidents and judging causes, not hunting through menus. exemONE AI Assistant cuts down repetitive lookups and screen switching, and connects monitoring, root cause analysis, and report writing into one flow, helping operations teams make decisions faster.
FAQ
Q1. Do I need to install a separate agent?
exemONE AI Assistant uses the data exemONE already collects, with no separate agent or additional collection. It broadly supports major commercial and open-source DBMSs, including Oracle, MySQL, PostgreSQL, SQL Server, Tibero, MongoDB, and SingleStore, and it is included in exemONE, our integrated monitoring solution.
Q2. Which monitoring targets are supported?
You can use it with the monitoring targets exemONE supports, including databases, applications, servers, Kubernetes, cloud, and networks. Answers use metrics and analysis perspectives suited to each target type.
Q3. What kinds of questions can I ask in natural language?
Ask in everyday language, for example "Show me recent Critical alerts," "Which service has the slowest response time?", "Analyze this screen," or "Summarize the current status as a report." The AI understands the intent, queries the right data, and takes you to the screen you need.
You no longer need to learn complex menus or know how to write queries.
Ask in your own words, and exemONE AI Assistant finds and shows you what you need.
👉🏻 Contact us
Related articles
