The control plane behind your private cloud
Operate private cloud infrastructure, applications, and AI from one workspace. Connect your own hardware or cloud accounts; Anchras coordinates lifecycle, policy, and audit while workloads stay under your control. Currently in closed beta.
Two current views of the control plane.
See the platform's shared operating surface and its governed AI domain in two silent walkthroughs.
Platform overview
Move from the operating dashboard through infrastructure and Kubernetes to a 120+ application catalog with sovereignty context visible before deployment.
Read the walkthrough
The workspace dashboard brings infrastructure, applications, and governed AI into one operating view. The walkthrough opens the infrastructure inventory, where operators can register on-premises capacity or connect supported cloud providers; moves to the Kubernetes cluster surface; and finishes in the application catalog. Catalog entries expose licensing, resource requirements, configuration, and an eight-objective sovereignty assessment with a SEAL level before deployment.
Governed AI
Inspect the AI control plane from gateway and model routing to versioned agents and the audit surface that records policy and runtime decisions.
Read the walkthrough
Anchras treats AI as a governed operating domain. The walkthrough starts with the AI overview, then shows OpenAI-compatible gateway access, provider and routing controls, the model registry, and versioned agent definitions with scoped tools and secrets. It closes on the AI audit surface, where operators can filter gateway, provider, model, guardrail, usage, memory, agent, workspace, swarm, local-inference, and tool-approval events.
Most teams face a false choice between a hyperscaler that owns their stack and a DIY build that owns their weekends.
Public cloud is fast to start and slow to leave. Egress fees, opaque pricing, shared-tenancy compliance theatre, and a jurisdiction you did not choose are the price of convenience. The exit interview gets harder every year. Build-it-yourself trades that dependency for a different one: a specialist team, on-call rotations, a hardware refresh cycle, and infrastructure work that competes with the thing you are actually trying to ship.
Anchras Platform is the third option. We package the operations work, the curated software, the networking, and the AI governance into one control plane, while you choose where workloads and data run. The control stays with you; the operational burden does not.
Three domains. One control plane.
The platform is organised around the work operators actually do. Private cloud is the infrastructure. Applications is the software you run on it. AI is the model access governed alongside everything else. Dashboard, CLI, Terraform provider, and an MCP server drive the same API.
Join the beta waitlist-
01
Private Cloud
Register Proxmox VE, Hyper-V, and XCP-ng capacity, install Anchras Node OS on greenfield hardware, or connect supported cloud accounts. Run VMs, containers, K3s and RKE2, or import EKS and AKS. Tailscale or Headscale keeps the private paths visible in the same operator surface.
-
02
Applications
Install software from a catalog of more than 120 curated applications, or bring your own Helm chart or Compose file. Each catalog app is scored with the Anchras Sovereignty Score — the European Commission's Cloud Sovereignty Framework applied per app, eight objectives each with a SEAL level, rolled into a 0–100 score — and the deploy wizard shows how that score changes for the hardware you place it on. Helm- or container-based; deployments land on a cluster, a fleet, or a single instance with a private tailnet endpoint by default, and go public through Tailscale Funnel only when you say so. Upgrade, roll back, expose, and audit from one page.
-
03
AI
Governed model access alongside the rest of the platform. An OpenAI-compatible gateway with a provider registry and your own credentials, routing with fallback, hard spend budgets, guardrails on prompts, outputs, and tool calls, and payload retention off by default. Versioned agents with a default-deny tool allowlist and approval-gated runs, swarms, pgvector-backed memory, and coding-agent workspaces on your tailnet. Deploy Ollama or vLLM from the catalog and it registers itself as a gateway provider. Every model call is attributed and audited.
The operating surface, in detail.
Anchras follows a workload from the capacity underneath it to the identity, network, data, policy, and audit trail around it.
-
Compute & Node OS
Bring Proxmox VE, Hyper-V, and XCP-ng hosts into one inventory, or install Anchras Node OS directly on greenfield hardware. Track capacity and health, manage VM and container workloads, connect cloud accounts, and make GPU accelerators available where workloads need them.
-
Kubernetes lifecycle
Provision K3s and RKE2 clusters on private infrastructure or import managed EKS and AKS clusters. Add, drain, and scale nodes, coordinate upgrades, and keep kubeconfig material encrypted.
-
Application delivery
Deploy from a catalog of more than 120 curated applications. Configuration forms are generated from each app definition; Helm-backed releases expose live logs and support upgrade, rollback, and removal from the same workflow.
-
Networks & endpoints
Connect a customer-owned Tailscale tailnet, use Anchras-managed Tailscale, or use Anchras-hosted Headscale. Manage enrolment, MagicDNS, tags, and common ACL rules without hiding the raw policy. Applications stay private by default; Funnel is an explicit public-exposure decision.
-
Storage & data services
Manage cloud object storage across S3, Azure Blob, and Hetzner Object Storage, alongside volume and snapshot inventory, TrueNAS-backed file services, and Postgres-first managed databases.
-
Cloud edge
Coordinate load balancers, public DNS zones, private application endpoints, and TLS from the control plane. Keep public exposure separate from the private routes used by operators and internal services.
-
Identity, secrets & audit
Organisations, teams, roles, invitations, SSO/OIDC, and scoped API keys establish who can act. KMS-backed secrets keep provider credentials and workload keys under organisation ownership; audit events record the resulting changes.
-
Operations & automation
Inspect health, capacity, metrics, logs, alerts, jobs, notifications, and investigations across the estate. Work through the dashboard, the platform-aware Copilot, generated OpenAPI documentation, the
anchrasCLI, or the core resources in the Terraform provider. -
Governed model access
Route applications through an OpenAI-compatible gateway using hosted providers, your own credentials, or approved local inference endpoints. Keep model availability, routing policy, guardrails, spend budgets, and usage attribution in one place.
-
Agents, memory & workspaces
Define versioned AI agents with scoped tools and secrets, run them asynchronously, and inspect each trace. Shared memory, coding workspaces, private AI applications, swarms, and AI audit all use the same organisation and policy boundaries as the infrastructure below them.
From capacity to governed workload in four steps.
-
01
Bring capacity
Register a supported hypervisor with the lightweight agent, install Anchras Node OS on bare metal, or connect a supported cloud account.
-
02
Establish the private fabric
Connect a customer-owned Tailscale tailnet, use Anchras-managed Tailscale, or use Anchras-hosted Headscale. Devices, services, policy, and live path telemetry stay visible together.
-
03
Choose the runtime
Provision VMs, containers, K3s, or RKE2; import EKS or AKS; then attach the storage, accelerators, endpoints, and policy the workload needs.
-
04
Run applications and AI
Deploy curated or customer applications, managed data services, and governed AI. Roles, secrets, budgets, network policy, telemetry, and audit follow the workload.
Sovereignty is the operating principle, not a marketing tag.
Anchras is Belgian and EU-grounded. GDPR is jurisdictional, not a checkbox; the control plane runs in the EU while workload placement stays an explicit customer choice, and a Data Processing Agreement is part of every paid engagement. Security controls are documented and open to review. Certifications and sector-specific frameworks (NEN 7510 for Dutch healthcare, PCI DSS for payment processing) are scoped per engagement rather than pre-claimed. The Sovereignty Score applies the European Commission's Cloud Sovereignty Framework to every catalog app across eight objectives, rolled up into a comparable SEAL level.