Deploy
Production-ready manifests, pushed to Git and delivered with Argo CD. Guided templates for common tools, on Kubernetes or OpenShift.
AI-powered Kubernetes operations platform
Deploy → Observe → Investigate → Optimize.
One platform for the whole operational loop. Connect a cluster in minutes, deploy through a guided wizard, and let KrevoPilot watch it — explaining real problems with evidence, recommending fixes, and turning wasted CPU and memory into an optimized GitOps pull request.
Privacy-filtered agent
GitOps-ready output
Kubernetes + OpenShift
prod-openshift-01 connected
payment-api is crashing repeatedly
Alert
CrashLoopBackOff
CrashLoopBackOff
Evidence
8 sources
8 sources
Root cause
OOMKilled
OOMKilled
Root cause identified · 92%
High memory usage causing OOM kills
Container exceeded its 512Mi limit after the latest image update.
The platform
One control plane that covers the entire operational workflow.
Most tools stop at one stage. KrevoPilot follows an application from the moment you deploy it to the moment it costs you too much — deploy it, watch it, explain it when it breaks, then right-size it.
1Connect a cluster in minutes with a read-only Helm agent
2Deploy an application through the guided wizard
3KrevoPilot monitors it and detects real problems
4Krevo AI explains the root cause with evidence and recommends a fix
5After a week, the cost optimizer finds wasted CPU and memory
6Generate an optimized GitOps pull request
Deploy → Observe → Investigate → Optimize
Four stages. One platform. No handoffs between tools.
Every stage feeds the next: what you deploy is what gets observed, what breaks is what gets investigated, and what runs is what gets optimized.
Observe
Dashboards across applications, projects, pods and workloads — metrics, events, health and alerts, grouped the way your teams think.
Investigate
Krevo AI correlates logs, events, deployments, configuration changes and metrics to explain the root cause — with the evidence behind it.
Optimize
Historical CPU, memory and storage analysis with weekly reports, AI recommendations, optional cloud cost estimation, and optimized YAML.
New Deployment
1 Choose cluster
2 Choose tool
3 Configure
4 Review & deploy
PostgreSQLRedisKafkaGrafanaArgo CDHarbor
Deployment builder
Generate production-ready Kubernetes and Argo CD packages.
Choose a cluster, select a private app or public tool, enter the required production settings, and let KrevoPilot generate the GitOps package.
- Guided templates — every field explains what it is and why it matters
- Raw Kubernetes YAML and Kustomize packages
- Argo CD Application files and Git paths
- Helm values for popular platform tools
- Kubernetes and OpenShift, with Krevo AI review before push
Observe
See every application the way your team actually organises them.
Projects, workloads and pods — not a wall of raw resources. Health, metrics, events and alerts in one place, so a developer can find their own service without learning the whole cluster.
- Applications grouped by project and namespace
- Live CPU, memory and restart trends per workload
- Events, health status and alerts in context
- Drill from a project down to a single pod
Applications
ecommerce · 6 workloads · Healthy
checkout-api · 4/4 pods · Healthy
payment-api · 1 restart · Warning
orders-db · 2/2 · Healthy
MetricsEventsHealthAlertsPodsLogs
AI investigation engine
Not just chat. A step-by-step incident investigator.
Krevo AI checks status, events, metrics, logs, recent changes, and deployment configuration before it gives an answer.
Evidence-first RCA
Correlates pod status, restarts, events, previous logs, resource limits, and recent deployment changes.
Hypothesis validation
Rules out image pulls, probes, DNS, PVCs, config, scheduling, and node pressure before naming the root cause.
Actionable fixes
Suggests commands, YAML changes, rollback options, and PR-ready remediation steps.
Optimize
After a week of real data, see exactly where your cluster wastes money.
KrevoPilot analyses historical CPU, memory and storage — comparing what every workload requests against what it actually uses — then hands you the fix as a pull request, not a chore.
Requested vs. used
Analyses historical CPU, memory and storage from the agent to expose the gap between what's allocated and what's consumed.
Waste, ranked
Surfaces the workloads holding the most over-allocated memory and CPU, ordered by the biggest saving — with weekly reports.
Right-size recommendations
Suggests safer requests from real peak usage, with optional cloud cost estimation once you add your own resource prices.
Optimized pull request
Turns the recommendation into optimized YAML and a GitOps pull request your team reviews — never applied automatically.
Live logs
Fetched on demand · nothing stored
Redacted inside your cluster
Optional: deploy Loki in one click
History stays in your cluster
Live by defaultLokiYour storage
Logs
Live logs by default. History on your terms, in your cluster.
Logs are the most sensitive data you own, so we do not hoard them. KrevoPilot fetches live logs on demand, redacts them inside your cluster before they travel, and stores nothing by default.
- Live pod logs on demand — no log warehouse, no per-GB bill
- Tokens, secrets and personal data masked before anything leaves
- Want history? Deploy Loki to your cluster in one click
- Your Loki, your disk, your retention and compliance boundary
Security & privacy
Designed for companies that care what leaves the cluster.
Secret-safe
Secret values are not collected. Metadata and references are enough for investigation.
Sanitized logs
Logs are filtered and redacted before any AI workflow uses them.
Read-focused agent
The agent sends operational snapshots; fixes still go through Git and review.
Audit trail
Track AI usage, admin actions, tenant activity, and investigation history.
Ready for MVP pilots
Deploy, observe, investigate, optimize — without stitching four tools together.
Start with one cluster and one agent. Connect in minutes, deploy your first application through the wizard, and grow into investigation and cost optimization when you're ready.