AI-native cloud platform · Built and run in the EU

The cloud platform built for
AI — end to end.

Compose complete, governed AI services from curated, versioned Products with standardized contracts. Your data boundary. Full operational insight.

Creating an organization is a request. Nothing is provisioned and nothing is billed until a human approves it.

EU operated

Platform operations and processing remain within the European operating boundary — infrastructure, data, and even the email we send you.

Dedicated customer environment

Your Resources run inside a dedicated customer environment with a controlled data boundary. Credentials and data do not cross it.

Auditable by design

Versions, changes, policy decisions and platform actions remain visible and traceable — a property of the architecture, not a promise in a document.

01 · Composable AI Resources

Describe the need.
The platform composes the solution.

A conversation becomes a versioned definition, and the Control Plane turns that definition into a running service.

Platform previewComposing a whole service from one definition is in development, and some Products above are announced rather than available. Products, Resources and Projects are available today.

1LLM chat
compose a service
Product owner
Build a secure health assistant for Frende. It should use our own documents, protect personal data and run with full traceability.
Varde Flyt assistant
I will compose a governed AI service from the Frende module, MCP behind a Context Gate, App Runtime, RAG over your approved sources, Rute with a Vern health policy and an approved AI model.Writing frende-health-assistant.json
2AI service definition
frende-health-assistant.jsonPlatform preview
1{
2 "apiVersion": "solutions.vardeflyt.no/v1",
3 "kind": "AIServiceDefinition",
4 "metadata": { "name": "frende-health-assistant" },
5 "spec": {
6 "products": {
7 "module": { "name": "frende-module" },
8 "connector": {
9 "product": "mcp@2.0.0",
10 "access": "context-gate"
11 },
12 "application": { "product": "app-runtime@0.1.0" },
13 "knowledge": {
14 "product": "rag@3.1.0",
15 "sources": ["approved-policies"],
16 "store": "document-database"
17 },
18 "gateway": { "product": "rute@2.3.0" },
19 "privacy": {
20 "product": "vern@1.8.0",
21 "policy": "pii-health"
22 },
23 "model": { "product": "ai-model@4.2.0" }
24 },
25 "policies": {
26 "dataBoundary": "dedicated",
27 "audit": "required",
28 "egress": "deny-by-default"
29 }
30 }
31}

Written by the assistant, not by hand. It is the desired state: which Products the service is made of, at which versions, how they connect, and the rules that hold.

3Running service
frende-health-assistant
Frende Module
PendingDeployingRunning
MCP + Context Gate
PendingDeployingRunning
App Runtime
PendingDeployingRunning
RAG
PendingDeployingRunning
Rute
PendingDeployingRunning
Vern
AI Model
PendingDeployingRunning

One service, in a dedicated environment.

One AI service, composed from Products. Each part runs as a Resource inside a Project of your organization — which is where access, policy and cost stay legible.

What the catalog curates

Clients

Apps people use directly.

AI services

Services that support applications, automated workflows, and system-to-system processes.

Knowledge & retrieval

Connects AI to trusted knowledge.

Integrations

Connects AI to tools and systems.

AI models

Deploy and run AI models.

Data & security

Protects sensitive data and AI workloads.

Announced: Vern — Protect sensitive data in AI workloads. Hosted Models — Build, customize and deploy your own AI models.

02 · Enterprise & AI governance by construction

Every AI answer.
Traceable and accountable.

Governance does not stop at the infrastructure. Versions, policy decisions and platform actions are already a matter of record; carrying that same discipline down to each individual answer is what the platform is being built toward.

ai-governance-recordPlatform preview
tracea7f3c1e9
serviceenterprise-health-assistant
modeleu-hosted-chat@2.1.0
context-gateallowed
sources3 approved documents
protection2 fields redacted
routingeu-only pool
groundednesscited
outcomeanswered

Model governance

A Resource runs a released version of a Product, and a released contract is immutable — it is replaced by a new version, never edited. What a Resource was running is a matter of record.

Context and data lineage

Retrieval runs over the sources your organization approved, inside your data boundary. The platform operates the service and is never in the prompt or response path.

Policy and guardrails

What a Resource may reach is declared in its contract and denied by default, and the platform validates that contract before anything is deployed. Inside that boundary the Product owns its own request path.

Human oversight and replay

Released definitions and versions are immutable, so the exact composition that produced an outcome can be inspected and stood up again.

03 · Agent-native infrastructure

Humans define the need.
Agents compose the service.

A Context Gate is a machine identity in your own organization — CI, automation, an agent — and it takes the same grant a person takes. It gives approved agents a safe, machine-readable way to use the platform, while the Control Plane validates and reconciles every change.

1Human

Describe

Goals, users, data sources and guardrails.

2Agent

Compose

Selects approved Products, versions and configuration — entering through a Context Gate.

3Declarative

Define

Produces a versioned definition of the AI service.

4Control Plane

Deploy

Validates, authorizes, resolves dependencies, reconciles and observes.

Consistent

The same contracts, every time.

Efficient

Less generated code, less variation.

Auditable

Every action and version recorded.

Controlled

A dedicated customer environment.

Build the AI service your enterprise actually needs.

01

Describe

Define the outcome, the users, the data boundaries and the governance you need.

02

Compose

Curated Products are assembled into a validated definition of one AI service.

03

Run

The Control Plane deploys and operates it inside your dedicated customer environment.

Ready to build?

Turn your next AI idea into an accountable service.