Skip to main content
Version: 1.0.0-beta

AI Provider Configuration

Configure AI model providers through the Admin UI. Switch between local (Ollama) and cloud (Gemini, OpenAI, Anthropic, DeepSeek) providers without code changes.

Accessing AI Models Settings​

Navigate to Administration → AI Models.

Adding a Provider​

Step 1 — Click "+ Add Model"​

Step 2 — Select Provider​

ProviderBest ForRequires Internet
OllamaOn-prem, air-gapped, data-sensitive environments❌
GeminiBest overall quality and speed✅
OpenAIGPT-4o for complex reasoning✅
AnthropicClaude for long-context analysis✅
DeepSeekCost-effective alternative✅
CustomAny OpenAI-compatible endpointDepends

Step 3 — Configure​

FieldRequiredDescription
Name✅Display name (e.g., "Gemini Flash")
Provider✅Select from list
Model ID✅Model identifier (e.g., gemini-2.0-flash)
API URL✅Endpoint URL
API KeyDependsRequired for cloud providers
Type✅llm, reasoning, or embedding
System PromptOptionalCustom system instructions
TemperatureOptionalCreativity level (0.0 – 1.0)
Max TokensOptionalMaximum response length

Step 4 — Test Connection​

Click Test Connection to verify:

  • ✅ API endpoint is reachable
  • ✅ API key is valid (for cloud providers)
  • ✅ Model responds to a test prompt
  • ✅ Response time is acceptable

Step 5 — Save​

Click Save. The provider is now available for use.

Provider-Specific Setup​

Ollama (Local / On-Prem)​

Provider: Ollama
API URL: http://<ollama-host>:11434
API Key: (leave empty)
Model ID: llama3.1:8b (or your preferred model)

No API key needed. Requires Ollama running locally with models downloaded.

Gemini​

Provider: Gemini
API URL: https://generativelanguage.googleapis.com
API Key: (your Google AI Studio key)
Model ID: gemini-2.0-flash

OpenAI​

Provider: OpenAI
API URL: https://api.openai.com/v1
API Key: (your OpenAI API key)
Model ID: gpt-4o

Anthropic​

Provider: Anthropic
API URL: https://api.anthropic.com
API Key: (your Anthropic API key)
Model ID: claude-sonnet-4-20250514

Security​

  • API keys are encrypted at rest using AES-256-GCM
  • Keys are encrypted using the SECRET_ENCRYPTION_KEY environment variable
  • After saving, keys are displayed as •••••{last 4 chars} — never shown in full
  • Keys are transmitted over TLS only
  • Key rotation: delete and re-add the provider with a new key

Switching Default Provider​

  1. Navigate to Administration → AI Models
  2. Click the ⭐ icon next to the desired provider
  3. The selected provider becomes the default for all AI operations
  4. No code changes needed — routing is automatic

On-Prem / Air-Gapped Deployment​

For environments without internet access:

  1. Install Ollama on a local server
  2. Download models offline: ollama pull llama3.1:8b
  3. Configure AIIA to point to the local Ollama instance
  4. All AI features work without internet connectivity