WatchThatSite
Bring Your Own Keys

Google Vertex AI

Configure Google Vertex AI service account credentials.

WatchThatSite integrates with Google Vertex AI, providing access to enterprise-grade LLMs with advanced safety and governance features.

Prerequisites

  • Google Cloud Platform project
  • Vertex AI API enabled
  • Service account with appropriate permissions

Set Up Service Account

1. Create Service Account

  1. Go to Google Cloud Console
  2. Select your project (create one if needed)
  3. Go to Service Accounts (in IAM & Admin)
  4. Click Create Service Account
  5. Name it (e.g., "watchthatsite-vertex")
  6. Click Create and Continue

2. Grant Permissions

  1. Click Grant this service account access to project
  2. Select roles:
    • Vertex AI User — Required for model access
    • Vertex AI Service Agent — For additional capabilities
  3. Click Continue then Done

3. Create Service Account Key

  1. Go to Service Accounts page
  2. Click your newly created service account
  3. Go to Keys tab
  4. Click Add Key > Create new key
  5. Choose JSON
  6. Click Create
  7. Download file (keep it secure)

Enable Vertex AI API

  1. Go to APIs & Services
  2. Click Enable APIs and Services
  3. Search for Vertex AI API
  4. Click Enable

Wait a few minutes for the API to be enabled.

Get Your Credentials

From the downloaded JSON service account file, you need:

  • Project ID: "project_id" field
  • Service Account JSON: The entire JSON content (as password field)

The file looks like:

{
  "type": "service_account",
  "project_id": "my-project-id",
  "private_key_id": "...",
  "private_key": "...",
  ...
}

Determine Your Region

Vertex AI is available in multiple regions:

  • us-central1 — US Central
  • us-west4 — US West
  • europe-west1 — Europe (Belgium)
  • europe-west4 — Europe (Netherlands)
  • asia-northeast1 — Asia (Tokyo)

Check Vertex AI regions for availability.

Add to WatchThatSite

  1. Go to Settings > AI Providers
  2. Click Add Provider Key
  3. Select Google Vertex AI from the provider dropdown
  4. Enter:
    • Project ID: From service account JSON
    • Region: AWS region (e.g., us-central1)
    • Service Account JSON: Full JSON content from key file
  5. Optionally select specific models
  6. Click Save

Latest Vertex AI models with tool calling:

  • Gemini 2.0 — Latest, most capable
  • Gemini 1.5 Pro — High-performance, 1M token context
  • Gemini 1.5 Flash — Fast with large context
  • Claude 3.5 Sonnet — Via Bedrock integration

Model availability varies by region.

All Available Models

Your account has access to all models supporting tool calling in your region:

  1. Go to Settings > AI Providers
  2. Click Add Provider Key or Edit existing
  3. Models dropdown shows all available models
  4. Select specific models to restrict, or leave empty for all

Pricing

Vertex AI charges per 1M tokens:

  • Input tokens: Standard rate per 1M tokens
  • Output tokens: Higher rate per 1M tokens
  • Varies by model: Different models have different pricing

Check Vertex AI Pricing for current rates.

Advantages

  • Enterprise governance: Advanced compliance and audit features
  • Model garden: Access to various model providers
  • Fine-tuning: Ability to fine-tune models
  • Monitoring: Built-in monitoring and evaluation tools
  • MLOps: Integration with Google Cloud's ML ops stack

Security Best Practices

  • Restrict key permissions: Use custom roles with minimal permissions
  • Rotate keys: Regenerate service account keys periodically
  • Use VPC-SC: Enable VPC Service Controls for network isolation
  • Audit logging: Enable Cloud Audit Logs
  • Separate keys: Use different service accounts for different applications

Troubleshooting

"Invalid Credentials" Error

  • Verify project ID matches service account JSON
  • Check entire JSON is pasted (check for truncation)
  • Ensure service account has Vertex AI User role

"Permission Denied" or "Forbidden"

  • Verify service account has proper roles
  • Check IAM binding for service account
  • Confirm Vertex AI API is enabled
  • Wait a few minutes if roles were just assigned

"Project Not Found"

  • Verify project ID is correct
  • Check project exists and you have access
  • Ensure project ID matches your service account JSON

"Model Not Found" or "Region Not Supported"

  • Verify region supports the model
  • Check model is enabled in your project
  • Some models may not be available in all regions
  • Use Vertex AI console to find available models per region

Authentication Failures

  • Regenerate service account key in GCP console
  • Download new JSON file
  • Paste complete JSON content
  • Check for invalid characters or truncation

Region Selection

Different regions have different model availability and latency:

  • Choose region closest to your users
  • Check model availability before selecting region
  • Some advanced features may only be available in specific regions

AWS Bedrock

Configure AWS Bedrock credentials for accessing foundational models.

Groq

Configure Groq API key for ultra-fast inference.

On this page

PrerequisitesSet Up Service Account1. Create Service Account2. Grant Permissions3. Create Service Account KeyEnable Vertex AI APIGet Your CredentialsDetermine Your RegionAdd to WatchThatSiteRecommended ModelsAll Available ModelsPricingAdvantagesSecurity Best PracticesTroubleshooting"Invalid Credentials" Error"Permission Denied" or "Forbidden""Project Not Found""Model Not Found" or "Region Not Supported"Authentication FailuresRegion Selection