Skip to main content
CloudThinker runs as a hosted service at app.cloudthinker.io, and most teams never need anything else. When some work has to happen inside your network, you have two options: Workers for individual folders and hosts, or a private deployment of the whole platform.

​
Pick a model

Early access. Dedicated, self-hosted, and air-gapped deployments are offered through a design-partner program. System requirements, supported topologies, pricing, upgrade paths, who runs operations, and how the model is provided without internet access are confirmed during onboarding, not published here.

​
Choose a deployment

CloudThinker Cloud

Start today on the hosted service, with nothing to install.

Cloud with Workers

Keep file and shell work on machines you control. In beta.

Dedicated cloud

Get a single-tenant instance, hosted by CloudThinker in a region you choose.

Bring your own cloud

Run self-hosted or air-gapped inside your own perimeter.

​
Self-hosted reference architectures

See how a self-hosted deployment is laid out on each platform before you request one.

AWS

Amazon EKS, Aurora, and ElastiCache in your AWS accounts.

Microsoft Azure

AKS, PostgreSQL Flexible Server, and Azure Managed Redis in your subscriptions.

Google Cloud

GKE, Cloud SQL, and Memorystore in your Google Cloud projects.

FPT Cloud

FPT Kubernetes Engine and managed databases in your FPT Cloud tenant.

​
What to ask for in a vendor review

Uptime commitments go by plan: see the Support row in Plans and pricing. These docs don’t publish the hosting region, subprocessor list, status page, or support response times for any model. Ask for those when you contact CloudThinker, so your review doesn’t stall on them later.

​
Try Workers before a private deployment

Workers are in beta: CLI command names and flags may change before general availability. A Worker often covers the need that prompts a self-hosting request, with nothing to install beyond the CLI. It’s a folder on your machine where CloudThinker reads files and runs commands, connecting outward over HTTPS with no inbound port. Your firewall only needs to allow outbound HTTPS (port 443) to app.cloudthinker.io, directly or through an HTTPS_PROXY. If the proxy re-signs HTTPS, set SSL_CERT_FILE to your company CA bundle. Installing and updating the CLI also reaches cloudthinker.io and github.com. If every row you care about says yes, start with Workers.

​
Request a private deployment

1

Collect your requirements

Put these in your first message so the reply can be specific:
  • Region or data residency rules
  • Whether your hosts have internet access
  • The identity provider you use for sign-in
  • Roughly how many people will use CloudThinker
  • The clouds and accounts you plan to connect
  • The LLM provider you want to use, if you have one (see BYOK)
  • Any compliance frameworks your review checks against
2

Contact CloudThinker

Send those requirements through cloudthinker.io/contact and ask for a dedicated, self-hosted, or air-gapped deployment.Success state: you have a reply that says whether your environment fits the program and what onboarding needs from you.
3

Evaluate in the hosted cloud meanwhile

While you wait, connect a non-production account in CloudThinker Cloud to test the agents on real work. See the Quickstart.

Self-Hosted Sandbox (Workers)

Run chats and Automations in a folder on your own machine.

Data protection

See how CloudThinker stores and protects your data.

Security overview

SSO, SCIM, roles, and Audit Logs, which also record commands run on Workers.

BYOK

Bring your own LLM provider credentials.