ADR-003: AI Provider Abstraction

Status

Accepted

Context

Wisdoverse Nexus needs to integrate with multiple AI providers:

• OpenAI (GPT-4, GPT-4o-mini)
• Anthropic (Claude)
• Future: Gemini, local models, custom providers

Each provider has:

• Different API formats
• Different authentication methods
• Different streaming protocols

Decision

Create a unified AIProvider trait with provider-specific implementations.

#[async_trait]
pub trait AIProvider: Send + Sync {
    fn name(&self) -> &'static str;
    async fn generate(&self, req: GenerateRequest) -> Result<GenerateResponse, ProviderError>;
    async fn generate_stream(&self, req: GenerateRequest) -> Result<ProviderStream, ProviderError>;
}

Use Provider Registry pattern for dynamic registration and selection.

Consequences

Positive

• Easy to add new providers
• Consistent interface for all AI operations
• Support for fallback/redundancy
• Easy testing with mock providers

Negative

• Abstraction may not capture provider-specific features
• Additional complexity layer

Mitigation

• Allow provider-specific options via metadata field
• Document provider-specific behavior clearly