Sveltos
Sveltos is a unified orchestration platform that lets platform engineering teams declaratively define and deploy Kubernetes add-ons, Helm charts, and policies across an entire fleet of clusters from a single management cluster. It uses standard Kubernetes CRDs like ClusterProfiles and Profiles with label selectors for smart targeting, and supports progressive rollouts with health-check gates. The platform has been tested at scale against 15,000 clusters.
- Developer Tools
- Software Only
- Artificial Intelligence
- AI Agents
- Enterprise Software
Funding
Founders
Founder details are not available yet.
Product
Problem
Managing configurations, add-ons, and policies across a large fleet of Kubernetes clusters is complex and error-prone when done manually or with fragmented tooling. Platform engineering teams struggle to maintain consistent state, stage rollouts safely, and ensure security across distributed environments, especially as cluster counts scale into the hundreds or thousands.
Solution
Sveltos provides a unified orchestration layer that lets teams declaratively define and deploy Kubernetes add-ons, Helm charts, and policies across an entire fleet from a single management cluster. Using standard Kubernetes CRDs like ClusterProfiles and Profiles, administrators define desired state once and use label selectors to automatically target the right clusters. The platform supports multiple sync modes—OneTime, Continuous, and ContinuousWithDriftDetection—ensuring clusters always match the declared intent. Sveltos also enables progressive rollouts with staged promotions across environments, job-based health checks that validate deployments, and a pull-based agent model for zero-trust security without opening inbound ports.
Target Audience
Platform engineering teams, DevOps/SRE organizations, and Kubernetes administrators managing multi-cluster environments at scale, particularly those running GitOps workflows or needing progressive delivery across production fleets.
Features
- Pull-mode agent architecture that connects outbound to the management plane, enabling zero-trust management without exposing API servers or opening inbound ports
- ClusterProfiles and Profiles CRDs for declarative definition of add-ons, Helm charts, and policies with label-selector-based smart targeting
- Progressive rollout and promotion engine using the ClusterPromotion CRD to stage deployments across dev, staging, and production with health-check gates
- Job-based health verification that runs smoke tests, synthetic transactions, and connectivity probes as Kubernetes Jobs inside managed clusters
- Native MCP (Model Context Protocol) Server that allows AI SRE tools to discover, query, and manage the Kubernetes fleet using natural language
- Multiple sync modes including OneTime, Continuous, and ContinuousWithDriftDetection for automated state reconciliation
- Push and pull distribution modes with a lightweight agent ensuring local cluster state matches global intent
- Proven scalability to 15,000 clusters as demonstrated in large-scale testing with vcluster