Skip to main content

LLM Configuration

ekai uses Large Language Models (LLMs) for AI-powered features including schema onboarding, ERD generation, BRD interviews, and artifact generation. Configure your LLM provider based on your organization's security and compliance requirements.

Owner Access Only

LLM Configuration is only accessible to account Owners. Admins and Users do not see this menu item.

Configuration Screen​

Navigate to LLM Configuration in the sidebar to configure your AI provider.

LLM Configuration screen showing provider options

Provider Types​

Public LLM APIs

Connect to hosted models via standard API keys.

ProviderModels
AnthropicClaude Sonnet 4.6
OpenAIGPT-5 mini
GoogleGemini 3.1 Pro

Use case: Quick setup, standard security requirements

Enterprise Cloud APIs

Use secure, private endpoints via your enterprise cloud accounts.

ProviderDescription
Google Vertex AIClaude & Gemini via Google Cloud
Azure OpenAIGPT models via Azure

Use case: Enterprise security, data residency, compliance

Snowflake Cortex LLMs

Coming Soon

Run models directly within Snowflake. Complete in-platform processing.

Use case: Enhanced security, Snowflake Native App deployments

Configuration Steps​

Public APIs​

  1. Navigate to LLM Configuration in the sidebar
  2. Select Public APIs
  3. Enter API keys for your chosen providers:
    • Anthropic API key
    • OpenAI API key
    • Google AI API key
  4. Click Test to verify connectivity
  5. Click Save

Enterprise Cloud (Google Vertex AI)​

  1. Select Enterprise Cloud
  2. Choose Google Cloud Platform tab
  3. Upload your Service Account JSON key or paste the content
  4. Click Test to verify connectivity
  5. Click Save
Service Account Requirements:
  • Vertex AI User role
  • Access to Claude and/or Gemini models
  • Appropriate project permissions

Enterprise Cloud (Azure OpenAI)​

  1. Select Enterprise Cloud
  2. Choose Azure tab
  3. Enter Azure OpenAI credentials:
    • Endpoint URL
    • API Key
    • Deployment names
  4. Click Test and Save

Model Usage by Task​

ekai automatically selects the optimal model for each task:

TaskPrimary ModelFallback
Schema OnboardingClaude Sonnet 4.6Gemini 3.1 Pro
ERD GenerationClaude Sonnet 4.6GPT-5 mini
BRD InterviewClaude Sonnet 4.6GPT-5 mini
Code GenerationClaude Sonnet 4.6GPT-5 mini
ValidationGemini 3.1 ProGPT-5 mini

Security Considerations​

What Goes to LLM Providers​

SentNot Sent
Schema metadata (table/column names)Raw data values
Statistical summariesIndividual records
Business requirement textSensitive business data
Generated code for reviewCredentials or secrets

Enterprise Cloud Benefits​

Why use Enterprise Cloud APIs?
  • Private cloud endpoints—no public internet
  • Enterprise security controls and audit logging
  • Data residency compliance (regional endpoints)
  • Your cloud provider's compliance certifications apply

Troubleshooting​

IssueSolution
Test failsVerify API key/credentials are correct and have required permissions
Model unavailableCheck your provider subscription includes the required models
Rate limitingEnterprise Cloud APIs have higher limits; consider upgrading
Timeout errorsCheck network connectivity to API endpoints

Next Steps​