Platform
Promote a channel the way you promote code.
Every project gets an isolated sandbox for free. Long-running staging and production environments run the identical engine image, behind gates you configure — so the thing you soaked is byte-for-byte the thing that goes live.
Environment model
Four kinds of environment.
Sandbox is where you build. Everything else costs money because it stays up and carries real traffic.
Sandbox
Free on every planEphemeral and isolated. Synthetic data only. Spins up on demand and shuts down after an idle timeout — 60 minutes on Free, 240 minutes on Team and Enterprise.
- Included on every plan
- Synthetic data only
- Idle TTL 60 / 240 minutes
Staging
Long-running, paidA long-running environment that stays up. Point it at test endpoints of the systems you integrate with, and soak a channel before it carries live traffic.
- Long-running
- PHI under BAA on paid plans
- Counts against your environment allowance
Production
Long-running, paidEach channel runs in its own dedicated task. Inbound allowlists are required, retention runs to 30 days on Team and 365 on Enterprise.
- Dedicated task per channel
- Inbound allowlist required
- Team and Enterprise only
Custom
EnterpriseAdditional named environments — a validation tier, a partner-specific stage, a disaster-recovery target. Up to ten on Enterprise.
- Up to 10 on Enterprise
- Named to match your process
- Same promotion machinery
Free includes 0 long-running environments, Team includes 2, Enterprise includes 20 plus 10 custom. See the full entitlements table.
Promotion
Gates, not hope.
A release moves through a timeline you can watch, with each gate recorded.
-
- Gate
- Connector preflight
- When it runs
- Before promotion
- What it does
- Checks that every connector in the channel can reach its counterparty from the target environment — DNS, TLS handshake, credentials.
-
- Gate
- Soak
- When it runs
- After deploy, before cutover
- What it does
- Runs the channel in the target environment for a configured period and watches for errors before the gate clears.
-
- Gate
- Approval
- When it runs
- Human gate
- What it does
- A named approver signs off in-app. The approval, the approver and the timestamp land in the audit event stream.
Release timeline
- 01
Build
The revision is packaged with a pinned engine image digest.
- 02
Preflight
Connectors are checked against the target environment.
- 03
Soak
The channel runs under observation for the configured window.
- 04
Live
Traffic cuts over; the previous revision drains rather than dropping in-flight work.
Networking
Addresses you can hand to a firewall team.
Hospital network teams need two things before they will open a path: a stable address and a documented protocol. Cloud gateways give you both, and dedicated addresses give you a stable source IP for the counterparty's own allowlist.
- Cloud listeners live at
*.gw.intucloud.aiwith TLS 1.2 or higher. - Shared address pool by default; dedicated ingress or egress address sets as a $40/mo add-on.
- Production listeners require an inbound allowlist — an open production endpoint is not a supported configuration.
- Dedicated gateway addresses are included on Enterprise.
Same engine everywhere
What you tested is what runs.
There is one engine. The sandbox where you iterate, the staging environment where you soak, the production task carrying live ADT traffic, and a self-hosted deployment in your own data centre all run the same binary.
That is what makes the export story credible: leaving IntuCloud does not mean porting to a different runtime.
- Engine images are pinned by digest from sandbox through production.
- The same open-source intu-dev binary runs in every environment, including on your own hardware.
- A promotion moves a revision, not a rebuild — no environment-specific packaging step.
- Hosted in AWS us-east-1 during the beta.
Explore the platform
Start building today.
Free plan, no credit card — design and test unlimited channels.