
CloudThinker Enterprise reference architecture on FPT Cloud
Reference design. Self-hosted deployments are in early access. Components, sizing, and supported services are confirmed during onboarding and can vary by release. FPT Cloud services are shown by function; the exact services are confirmed with FPT Cloud for each deployment. See Deployment options.
Layers
How a request flows
- A user opens the web console over the private connection. The virtual router routes the request to the internal load balancer.
- The API checks the user’s role and queues the task for an agent worker.
- CloudThinker plans each step with the model you configured. Reads run, and every write goes through Approvals, which can wait for a person or refuse it.
- Commands run in a sandbox on the sandbox worker group. Connections reach on-premises systems through the private connection, and other systems through read-only accounts you create.
- Data, secrets, and logs stay in your environment. Backups are copied to the recovery region.
Design principles
What you provide
- An FPT Cloud tenant with a dedicated project for CloudThinker.
- A private connection (leased line or VPN) from your data centre.
- An identity provider that supports SAML 2.0 or OIDC. See Single sign-on.
- An FPT AI Factory account with approved models.
Related
Deployment options
Compare the hosted cloud, Workers, and private deployments.
Security and trust
Answer a vendor security review.
Approvals
Control which agent writes run, wait, or are refused.
BYOK
Bring your own LLM provider credentials.