Integrations Overview

Beluga AI is designed to work with the systems you already use. These integration guides cover connecting the framework to LLM providers, vector databases, cloud infrastructure, observability platforms, and communication channels. Each guide includes configuration, authentication, working code examples, and troubleshooting for the specific service.

The guides range from drop-in provider registrations (one import line) to custom loaders and retrievers that you build to match your infrastructure.

Integration Categories

Category	Guides
Agents & Tools	Tool registry, MCP integration, API bridges, testing
LLM Providers	OpenAI, Anthropic, Google, Bedrock, and 18 more
Embeddings	OpenAI, Cohere, Ollama, Jina, Voyage
Data & Storage	Vector stores, document loaders, S3, MongoDB, Redis, Elasticsearch
Voice	STT, TTS, S2S, VAD, transport, session management
Infrastructure	Kubernetes, HashiCorp Vault, NATS, Auth0
Observability	Langfuse, LangSmith, Datadog, Phoenix, Opik
Messaging	Slack, Twilio, webhooks
Prompts & Schema	LangChain Hub, filesystem templates, JSON Schema
Safety & Compliance	JSON reporting, ethical API filters

Integration Pattern

Beluga AI uses a registry-based architecture where providers auto-register via Go’s init() mechanism. This means most integrations follow the same three steps — import, configure, create — regardless of the underlying service:

// 1. Import the provider (auto-registers via init())
import _ "github.com/lookatitude/beluga-ai/llm/providers/openai"

// 2. Configure
cfg := config.ProviderConfig{
    APIKey: os.Getenv("OPENAI_API_KEY"),
    Model:  "gpt-4o",
}

// 3. Create and use
model, err := llm.New("openai", cfg)