Skip to content

OPENMARU CogentAI

A screen that took 20 seconds —
ask what is wrong and it answers

An AI agent that adds internal document search (RAG) and system integration (MCP) to a large language model (LLM). Inside your performance management and Kubernetes operations screens it tells you the cause of an incident first, reads slow SQL and explains it, and writes the analysis report for you. No external AI service is called. A small model runs on an in-house GPU server, so it works even on air-gapped networks.

Watch AI demosFind a product guide
CogentAI · APM · 2 min 37 sec

What changes when you use AI

OPENMARU CogentAI is not a separate program you launch. It works inside the APM, COP and Observability screens you already use — reading the metrics a person used to read, and telling you the cause. Most demos below run about a minute.

CogentAI · APM59 sec

Detects an error and tells you the likely cause with it

CogentAI · APM1 min

Reads the SQL execution plan and explains what is wrong in plain words

CogentAI · APM1 min 18 sec

Gathers metrics and writes them up as a report you can export

CogentAI · COP34 sec

Ask instead of memorising commands, and it reports cluster state

COP2 min 6 sec

Turns the several steps of rolling back a failed deploy into one

Observability29 min 36 sec

Finds the cause in metrics, logs and traces collected without code changes

CogentAI · COP1 min 12 sec

Run a build and check its logs by talking to it

CogentAI · APM2 min 37 sec

Follow a delay a user reported from intake through to its cause

More: Cloud Native TV · Kubernetes TV

Why OPENMARU

Not a feature list — these are things we actually built recently. Each one links to the manual chapter that explains it.

Less work to install and instrument

Kubernetes attaches the APM agent for youNEW 새 항목APM Operator · Automatic instrumentationNo per-language agent to wire up, whatever you built it in — Java, Go, Python, Node.js, .NET and PHP work the same wayObservability · eBPF zero instrumentation

Give it source, get a build and a deploy

Build a container image from source alone, with no DockerfileCOP · S2I builds (Source-to-Image)Set up a build once, then build with a single buttonCOP · Building from the Console

Use less hardware

Several pods share a single GPUCOP · Devices (GPU sharing)Scale up at a set time, before the load arrivesCOP · Autoscaling (CronHPA)Scale on application metrics rather than CPUNEW 새 항목APM Operator · Autoscaling (APM HPA)One small model is enough — AI even in an air-gapped networkAPM · About CogentAI (one GPU server)

See only what you need, and keep it as a report

See Kubernetes and your applications on one screenObservability · APM dashboardBuild a dashboard from just the charts you wantObservability · Custom dashboardsSee where the current trend leads, drawn on the chartObservability · Chart forecastWork out the resources you will need nextObservability · SRE report (forecasting)Gather metrics into a PDF or Excel reportObservability · Custom reports

Find the cause quickly

Read the current state from 16 chart shapesAPM · Diagnosing with T-Map patternsNarrow a slow transaction down to 19 known patternsAPM · Finding the cause of a slow transactionAI reads hundreds of threads and points to the common cause and deadlocksNEW 새 항목APM · Finding causes with AI (thread dumps)See the web server in front of the WAS on the same screenAPM · Web server (performance · workers)AI reads a slow query’s execution plan the way a DBA wouldNEW 새 항목APM · AI query diagnosisAlgorithms point out the values outside the normal rangeObservability · Chart anomaly detectionLine an incident up against the deploy that may have caused itObservability · Timeline map

Check security, and keep the record

Check cluster settings against the CIS benchmarkCOP · CIS benchmarkSee workload security risks ranked by severityObservability · RisksDetect HTTP attack patterns as they happenObservability · Security attack detectionKeep an audit record of who did whatObservability · Audit logsControl which resources each person can seeCOP · Access control (RBAC)

What are you trying to do?

You do not need to know the product name. Pick what you are trying to do, or the symptom you are seeing, and we will take you straight there.

I am trying to do this

Step-by-step from the beginning

I want to deploy an application to KubernetesCOP · Choosing a build methodI want to automate deployment with GitOpsCOP · Deploying with ArgoCDI want pods to scale automatically with loadCOP · AutoscalingI want different people to see different resourcesCOP · Access control (RBAC)I want to try APM for the first timeAPM · 5-minute quick startI want to ask AI what caused an incidentNEW 새 항목APM · Diagnosing causes with AII want to be alerted when something failsAPM · Creating an alert policyI want to start monitoring without changing codeObservability · Installation

I have this problem

From symptom to cause

A pod never moves past PendingCOP · StorageI cannot reach the service I createdCOP · NetworkingSomething is wrong with the cluster and I do not know where to startCOP · Cluster diagnosticsThe screen is slow and I cannot tell what is slowAPM · Finding the cause of a slow transactionMemory keeps growingAPM · Checking for memory leaksWe are running out of DB connectionsAPM · Diagnosing DB connection pool exhaustionThere are too many alerts to readAPM · Reducing alert noiseThe agent will not connect, or no data arrivesAPM · Checking agent status

Which product are you using?

Pick the product you use and we will take you to its user guide.

Docs people look for alongside

APM Installation & Integration GuideAgent · Kubernetes · Events · REST APIAPM Release NotesChanges by versionLearn JBoss EAPOnline technical book in Korean — 22 chaptersKO