PortvMind Cloud · Agent Cloud · Coming soon

Run AI agents on your own container contract.

Agent Cloud is a coming-soon platform for running AI agents on PortvMind Cloud through Service and Job flows. You bring the image, command, port, secrets, volume, rollout and rollback.

Agent Cloud orchestration sceneA diagram of how an agent workload defined in the console connects to Service and Job resources, secret and volume bindings, external access through the gateway, and the log and metric flow.CONSOLE / PROJECTOrganisation and projectquota · work contextContainer contractimage · command · portRollout and rollbackrevision historySecret fieldswrite-only · never returnedVolume and snapshotpersistent data directoryAGENT RUNTIMESERVICEreplica · internal address · proberollingUpdateJOBcommand · time limitOBSERVABILITYlog · CPU/RAM · searchGATEWAYHTTPS accessdomain · portHealth signalstartup · readinessROADMAPModel layerplannedcapabilities
Resources defined in the console, the Service and Job lifecycle, gateway access and the observability flow. Dashed lines show secret and volume bindings.

Overview · Coming soon

What is Agent Cloud and what will it offer?

Agent Cloud is a coming-soon platform for running agent workloads through the container contract on PortvMind Cloud. You define your image, command, port, secrets and volume, then set up Service and Job flows.

Hermes Agent and custom container workloads will arrive with Agent Cloud. OpenClaw and n8n are planned capabilities. Interested teams can contact us.

Teams · Coming soon

Turn an agent workload into an operable cloud service.

The same infrastructure serves the visibility needs of business, automation and platform teams.

Business teams

Keep the agent's data and tool access as visible as the work it performs.

Sets a secure operating boundary for an approved agent workload.

AI and automation teams

Run Hermes, OpenClaw, n8n or your own container workload on the same deployment contract.

Image, persistent data, secrets, external access and update requirements are set in the deployment plan.

Platform and developer teams

Manage deployment, networking, observability and rollback together through the console.

The agent runs as an observable, reversible cloud workload.

Three product layers · Coming soon

Agent Cloud will sit in the middle of three layers.

The infrastructure layer is in production. Agent Cloud will soon run on it through the container contract. The model and agent gateway layer covers planned capabilities.

Layer 1 · Available

Infrastructure layer

Compute, network, storage, backup and Kubernetes: the PortvMind Cloud services already in production. Agent Cloud will run on top of them.

  • Compute and GPU resource profiles
  • Router, Load Balancer, Floating IP
  • Block Storage, Object Storage, Backup
Layer 2 · Coming soon

Console and container contract

The agent workload is defined by this contract: image, command, port, replica, probe, secret, volume, rollout and rollback. This is what Agent Cloud will offer.

  • Service and Job resources
  • Write-only secret fields
  • Logs, search and CPU/RAM metrics
Layer 3 · Planned

Model and agent gateway layer

Model resources, an OpenAI-compatible endpoint, scaling and the catalog experience are planned capabilities.

  • Model and ModelCache resources
  • vLLM and OpenAI-compatible Endpoint
  • Agent gateway and catalog

Console flows · 6 · Coming soon

Agent Cloud will offer six console flows.

The flows cover Service and Job, a private registry, secrets, HTTPS, logs, CPU/RAM metrics, volume, snapshot and rollback.

Coming soon

Project and work context

Manage agent resources in the correct organisation and project boundary.

  1. Select the organisation and project.
  2. Select the resource quota and work context.
  3. Continue to the Service or Job flow.

Coming soon

Publish a service from a container image

Define a persistent Service with image, resource, replica and port settings.

  1. Enter the container image and resource profile.
  2. Enter replica, command and port settings.
  3. Observe Service status and the internal endpoint.

Coming soon

Private registry and write-only secrets

Control image access with project policy and keys with write-only secret fields.

  1. Define the registry server and allowed image paths.
  2. Store the credential and model key as secrets.
  3. After saving, the secret value does not appear in the response.

Coming soon

HTTPS, health probes and rollout

Handle external access, health signals and controlled updates in one Service definition.

  1. Define the gateway port and HTTPS hostname.
  2. Add startup, readiness and liveness probes.
  3. Set RollingUpdate limits and publish.

Coming soon

Logs, search and CPU/RAM metrics

Observe the running agent service through live logs, history and resource use.

  1. Open live pod logs.
  2. Search history by time range and text.
  3. Compare CPU/RAM use with allocated resources.

Coming soon

Job, volume, snapshot and rollback

Manage one-off work, persistent data and rollback in one lifecycle.

  1. Define Job command, timeout and retry policy.
  2. Mount a volume and take a snapshot.
  3. Roll back a previous Service revision as a new version.

Agent catalog · Coming soon

Workloads coming with Agent Cloud.

Hermes Agent and custom container workloads will ship with Agent Cloud. OpenClaw and n8n are planned capabilities.

Coming soon

Hermes Agent

An open-source, container-based agent published by Nous Research.

Maps a persistent data directory, gateway and model key to a controlled Service flow.

Planned

OpenClaw

An open-source, self-hosted gateway that connects messaging channels to AI agents.

Planned scope: official image, persistent directories, gateway token, network isolation and sandbox.

Planned

n8n

A self-hostable automation tool that connects API-enabled applications and AI steps through visual workflows.

Planned scope: persistent data, webhook access and credential separation.

Coming soon

Custom agent container

An agent workload with its own image, command, port, secret and volume requirements.

Runs on the container contract with your own image, command, port, secret and volume definition.

Infrastructure · Coming soon

Agent workloads run on infrastructure located in Türkiye.

The PortvMind Cloud data center layer is the physical infrastructure under GPU and agent workloads.

AI and data center layers in PortvMind Cloud infrastructure

Planned capabilities · Coming soon

Capabilities planned alongside Agent Cloud.

These are planned capabilities on the Agent Cloud roadmap.

Contact · Coming soon

Agent Cloud is coming soon.

Interested teams can contact us. In the conversation we clarify the container contract, data and access requirements and rollback expectations together.

What to bring to the consultation

  • The agent’s container image, command and port requirements
  • Model key and credential storage expectations
  • Persistent data volume, snapshots and retention period
  • External access, domain and health-signal requirements
  • Acceptable interruption and rollback approach

Related services · Coming soon

Compute

The agent service’s CPU/RAM profile and runtime are defined by this service.

VKE (Kubernetes)

The orchestration layer beneath Service, Job, replica and rollout behaviour.

Block Storage

The agent’s persistent data directory, volume mounting and snapshot flow.

Frequently asked questions · Coming soon

Agentic AI and Agent Cloud

What Agent Cloud will offer, and which capabilities are planned.

What is agentic AI?

Agentic AI describes software workloads that break a goal into steps, use tools or APIs and check results instead of producing a single response. In production, that behavior needs explicit access, data, cost and rollback boundaries.

Is Agent Cloud on PortvMind Cloud a managed agent platform?

Agent Cloud is a coming-soon platform for running container-based agents on existing cloud services. A one-click catalog, managed model endpoint and agent gateway are planned capabilities.

Do Hermes, OpenClaw and n8n have the same status?

No. A deployment guide based on the official Hermes Docker contract is published. OpenClaw and n8n are planned capabilities.

How are agent secrets and external access handled?

API keys are stored in write-only secret fields, while external access is limited through HTTPS, health probes and an explicit port scope. Dashboards and APIs are not exposed to the internet without authentication.

Agent Cloud is coming soon.

It will be available soon; interested teams can contact us.

Scope and terms

Agent Cloud scope: