Skip to the operations

AI

Vertex AI in a decision flow.

3 Vertex AI operations a decision flow can call directly, with the credentials your own contract issued. The response is data the rest of the flow reads, branches on, and keeps in the trace.

Category
AI
Type
Integration
Authentication
Google service account
Test environment
Production host only

Who they are

Google Cloud's machine learning platform, including the Gemini models.

Vertex AI is where Google Cloud sells access to its own models, including Gemini, alongside tooling for training and serving custom ones. It is the same model family as the Gemini Developer API, reached through a Google Cloud project instead of an API key, which brings that project's data residency, network controls and billing. Enterprises usually choose it for those controls rather than for the models.

What a flow can call

3 operations, each one a step you can place on the canvas.

  1. POST/v1/projects/$data.gcp_project_id/locations/global/publishers/google/models/$data.vertex_model:generateContent

    Ask a model

    One request, one answer. Both the project and the model ride in the path, so a single connection serves several projects and a flow can move models without a new node.

  2. POST/v1/projects/$data.gcp_project_id/locations/global/publishers/google/models/$data.vertex_model:generateContent

    Extract structured JSON

    The same endpoint constrained to a schema, so the reply parses without a regex.

  3. POST/v1/projects/$data.gcp_project_id/locations/global/publishers/google/models/$data.vertex_model:countTokens

    Count tokens before sending

    Returns the token count for a request without running it.

Where it sits in the decision

AI calls have a natural place in a flow.

A model call is for the judgment a table cannot express — reading a document, summarising a file, weighing a narrative. Because the output is not deterministic, a flow constrains it with a schema and keeps the prompt and the response in the trace, so the reasoning behind a referral stays as inspectable as a rule.

Whatever Vertex AI returns is part of the run, so it is part of the record. When someone asks months later why an applicant was declined, the answer cites what came back at the time rather than re-fetching from a service whose answer has since changed.

  • 01Add Vertex AI as a connection authenticating with a Google service account.
  • 02Vertex AI has one host for both environments, so guard test runs with your own credentials and limits.
  • 03Place a Connection node and pick an operation — “Ask a model” is usually the first one a flow needs.
  • 04Map the response into the fields your rules read, then test the whole path before it carries live traffic.

Common questions

Using Vertex AI in a flow.

How do I connect Vertex AI to a decision flow?

Add Vertex AI as a connection in your workspace authenticating with a Google service account, with the credentials your own contract issued — ArboRule calls the provider as you, and never holds a contract on your behalf. Once the connection exists, any flow in the workspace can place a Connection node and choose one of its operations. The credentials live on the connection, not in the flow, so a policy owner can use Vertex AI in a decision without ever seeing the secret.

Can I test Vertex AI without touching production?

Vertex AI exposes one host for both environments, so there is no separate sandbox to point at. Test runs still execute in Sandbox and are recorded separately in decision history, but the call goes to the same place as production — so guard it with your own test credentials, rate limits, or data.

What can a flow call on Vertex AI?

3 operations: “Ask a model”, “Extract structured JSON”, “Count tokens before sending”. Each one is a step you place on the canvas and map into the fields your rules read, and most flows start with “Ask a model”. The list comes from the same manifest the engine uses to make the call, so this page cannot describe an operation the product does not have.

Where in a decision should Vertex AI be called?

A model call is for the judgment a table cannot express — reading a document, summarising a file, weighing a narrative. Because the output is not deterministic, a flow constrains it with a schema and keeps the prompt and the response in the trace, so the reasoning behind a referral stays as inspectable as a rule.

Also in AI

Ready when you are

Wire Vertex AI into a real decision.

Build the flow in Sandbox, connect your account, and watch the decision pull what it needs before it answers.

Read the docs