# Keep Belong AI Deploy Access Private

Source: https://belong.net/blog/keep-belong-ai-deploy-access-private
Published: 2026-11-11T04:13:00.396Z
Updated: 2026-11-11T04:13:04.186Z

Set up Belong AI Deploy Private Access so approved devices can reach an agent service without exposing its ports publicly.

Keep Belong AI Deploy Access Private

What it does

Belong CheckIn teams that deploy an AI agent can use Belong AI Deploy Private Access to limit access to approved devices. The capability is for operators who want to reach an agent’s management interface, SSH, and agent traffic through a private network without opening those ports to the public internet.

The configured service creates a private network between the agent server and the user’s devices. The server receives a private network IP, and connected devices receive their own addresses on the same network. They can then communicate directly as though they were on one local network.

The setup keeps management interface, gateway, and agent ports private. A separate HTTPS entrypoint can still be added intentionally, but it is not required for direct private access. If public webhook ingress is used, the documentation says it should run through one shared HTTPS entrypoint rather than exposing application ports directly.

Traffic over the private network is encrypted end to end, and management interface access is limited to devices on the same private network. This makes Private Access a defined path for reaching the deployment while keeping its working ports away from the public internet.

How it works

- Create a private-network account. Sign up with the documented private-network provider using a supported identity provider. The documentation notes that personal use is free for up to 100 devices.

- Generate an auth key. In the provider’s key settings, generate a key for the Belong AI Deploy wizard. Configure it as reusable, not ephemeral, and leave the default expiration. Copy the generated key, which begins with tskey-auth-.

- Add the key in the deployment wizard. Paste the auth key into the wizard’s private-network screen and select Verify. A successful check displays Private network ready.

- Finish the deployment. The agent server joins the private network and receives a private network IP. Its gateway and agent ports remain private unless a separate HTTPS entrypoint is intentionally configured.

- Install the private-network client on each approved device. The documentation lists support for iPhone, iPad, Android, Mac, Windows, and Linux. Install it on every device that needs to reach the management interface.

- Sign in with the same account. Open the private-network client on the approved device and use the same account associated with the private network. This places the device and agent server on the same private network.

- Open the management interface address. In a browser on the connected device, visit the documented private-network address on port 3000, replacing the placeholder with the server’s private network IP. That address opens the management interface through the private network.

This flow avoids opening the management interface, SSH, or agent traffic to the public internet. The server and approved devices communicate through their private network addresses, while any optional public webhook ingress remains separate.

Who it’s for and when to use it

Private Access is for teams deploying an AI agent with Belong AI Deploy and deciding how administrators should reach its management interface and server services. It is appropriate when access should be limited to devices that have joined the team’s private network.

Use it when operators need the management interface but do not want to expose port 3000 publicly. The same private-network approach applies to SSH and agent traffic: the documentation states that no public ports are required for these paths.

The setup also separates private administration from optional public ingress. Gateway and agent ports remain private by default. If the deployment needs inbound webhooks, those can use one shared HTTPS entrypoint under the operator’s control rather than making each application port public.

Each device that needs management interface access must run the private-network client and sign in to the same network. Access is therefore based on network membership rather than a public management interface address. The documentation lists mobile and desktop platforms, allowing an approved operator to use a connected phone, tablet, or computer.

At the event

Consider a venue operator who needs to check an agent’s management interface from a laptop during an event. The laptop has the private-network client installed and is signed in to the same account as the deployed server. The operator opens the browser and enters the server’s private network IP with port 3000.

The laptop and server communicate through the private network. The management interface port is not opened to the public internet, and management interface access remains limited to devices on that network. If another operator needs access from a different approved device, that device must also have the private-network client installed and use the same network.

If the deployment separately receives public webhooks, those do not require the management interface or application ports to become public. The documented pattern is a single, optional HTTPS entrypoint controlled by the team, while direct management access continues over the private network.

Get started

Follow the full Private Access documentation while configuring the deployment. It covers account creation, auth-key settings, verification in the wizard, device installation, the management interface address, the private-network model, and security details.

You can also open Belong CheckIn for Belong’s event experience. For the deployment itself, complete the private-network screen in the Belong AI Deploy wizard, verify the private-network auth key, and install the private-network client on every approved device before attempting to reach the management interface.

Keep the management interface, SSH, gateway, and agent ports private. Add a public HTTPS entrypoint only when the deployment intentionally needs public webhook ingress, and keep that ingress separate from the application ports.

FAQ

Does Private Access require opening the management interface port publicly?

No. Approved devices reach the management interface at the server’s private network IP on port 3000, and the documentation states that management interface, SSH, and agent traffic do not require ports to be opened to the public internet.

Which devices can access the private management interface?

Devices on the same private network can access it. The setup guide lists iPhone, iPad, Android, Mac, Windows, and Linux, with the private-network client installed and signed in to the same account.

Can a deployment still receive public webhooks?

Yes. Public webhook ingress is optional and should use one shared HTTPS entrypoint controlled by the operator. The agent’s application and gateway ports should not be exposed directly for that purpose.
