Runs on AWS, Azure and Google Cloud. Service by service, here is what each one has.
Pick a cloud, or leave it open and Intentric chooses. Vendor support is the easiest thing in this industry to overstate, so this page lists it as a table: each service, and whether it is available on each cloud.
Which service runs where.
Every service has the same rules, approval and verification wherever it runs. Where a cloud cannot support a service, a request for it is refused up front, with the reason, and never at apply time.
| Service | AWS | Azure | Google Cloud |
|---|---|---|---|
| Databases and data | |||
| PostgreSQL database | Available | Available | Available |
| MySQL database | Available | Not available | Available |
| SQL Server database | Available | Not available | Not available |
| NoSQL table | Available | Available | Available |
| Cache | Available | Available | Available |
| Data warehouse | Available | Not available | Not available |
| Compute | |||
| Container service | Available | Available | Available |
| Function | Available | Not available | Available |
| Scheduled job | Available | Available | Available |
| Virtual machine | Available | Available | Available |
| Kubernetes cluster | Available | Not available | Available |
| Kubernetes application | Available | Not available | Available |
| Container registry | Available | Available | Available |
| AI agent | Available | Available | Available |
| Storage | |||
| Object storage | Available | Available | Available |
| File share | Available | Available | Available |
| Secret | Available | Available | Available |
| Networking and delivery | |||
| Private network | Available | Available | Available |
| Load balancer | Available | Available | Available |
| CDN distribution | Available | Available | Available |
| DNS zone | Available | Available | Available |
| Private DNS zone | Available | Available | Available |
| Private endpoint | Available | Not available | Available |
| Site-to-site VPN | Available | Not available | Not available |
| Web application firewall | Available | Available | Available |
| API gateway | Available | Not available | Not available |
| TLS certificate | Available | Not available | Available |
| Security, identity and operations | |||
| Encryption key | Available | Available | Not available |
| User sign-in directory | Available | Not available | Not available |
| Pipeline roleOn Google Cloud for one container service. | Available | Not available | Available |
| Alert | Available | Not available | Available |
| BudgetA notice, never a cap. | Available | Available | Not available |
| Backup plan | Available | Not available | Not available |
| Audit log export | Available | Not available | Not available |
| Messaging and email | |||
| Message queue | Available | Available | Available |
| Message topic | Available | Available | Available |
| Transactional email | Available | Available | Not available |
Limits beyond the table: on Azure, a container or an AI agent cannot yet be connected to your other services. On Google Cloud, an AI agent cannot yet be connected to a database or storage, the content delivery network serves storage only, and the load balancer is internet-facing only.
How each cloud is linked, and what is distinctive.
You link an account, a subscription or a project once per environment. Intentric never holds a standing key to it.
Amazon Web Services
The broadest coverage: every service in the table above is available. You link an account by deploying a stack in it, and the permissions it grants are built from what the generated infrastructure creates and removes.
The sandbox runs the whole loop against a simulated AWS, so you can judge the rules, the approval and the verification before you connect anything.
Microsoft Azure
You delegate access to a subscription through a delegation you can withdraw. Access is scoped per environment, to one resource group and one set of capabilities, rather than per job.
Certificates cannot be created on Azure, and a container or an AI agent there cannot be connected to your other services. Both are refused up front, not at apply time.
Google Cloud
You connect one project per environment through workload identity federation, and Intentric never creates or holds a key to it.
Google Cloud has no transactional email service, so there is none there. An AI agent there cannot yet be connected to your database or storage, and a request that needs it is refused up front.
If you do not name a cloud, Intentric picks one on price and on where you already have an account, and shows you the scoring — including what the other clouds would have cost and why any of them were ruled out. It does not look at compliance, latency or quota, and it will not pretend to.
OpenTofu or Pulumi: both are a choice you can make.
Intentric produces standard infrastructure code, so the tooling your team already uses to store, review and run it keeps working exactly as it does now.
OpenTofu
The default engine. Standard, inspectable, and runnable by your team without us.
Terraform
The output is standard HCL, written against providers both tools use, so teams standardized on Terraform can read it and carry it on.
Pulumi
A deployment, a request or your build defaults can choose it. Every plan then renders, is checked, applies and is verified through Pulumi, and ships as a Pulumi.yaml you can run with the stock Pulumi CLI.
This sits upstream of what you have.
If your team already runs orchestration, state management or a review pipeline for infrastructure, none of it is wasted. Intentric produces the thing those tools were built to run — it does not compete with them for the same job.
- Bind a definition to a repository on your git host: a change there can be proposed as a plan automatically, and still goes through the same approval
- Your review and deployment pipelines continue to work on standard code
- Sign-in with Google, Microsoft, GitHub, GitLab or Bitbucket, single sign-on through your own identity provider, and user provisioning from Microsoft Entra ID or Okta
- Machine access for your own automation, held to the same rules as a person
- Email, and Slack and Microsoft Teams channels, for what needs a person and how it turned out, alongside each member's in-app inbox
- Signed, retried webhooks for the events your systems care about, and your record exported as CSV or JSON
You do not start from an empty account.
Existing infrastructure stays exactly where it is. You choose what comes under management and in what order — usually starting with whatever is least frightening to hand over first.
- Adopting something is a change like any other: an elevated approver signs off, and a backup of its state is taken first
- Terraform you already wrote can be imported, from the command line or the console, as a proposed definition with a report of anything it could not carry over — see for engineers
- Environments stay separate, with their own parameters and their own stakes
- Stop at any point — what has been produced is standard code and stays yours
Judge the loop before you connect a cloud.
The sandbox runs the whole path against a simulated AWS, with no card and no cloud account. When you are ready, sign in and link your own.