AI & the gateway

wess.dev is built for AI apps. Attach a model provider to your app and we inject a provider-agnostic gateway — your code calls one endpoint, and the owner can switch between Anthropic, OpenAI, Ollama (server or cloud), or any OpenAI-compatible endpoint without touching your app. Your app never holds a raw provider key.

Attach a provider

wess ai myapp anthropic            # or: openai, ollama, or any configured provider
wess deploy myapp --image ...      # redeploy to inject the gateway env

The owner configures providers once in the console (Settings → AI providers): an Anthropic or OpenAI key, or an Ollama base URL (local server or Ollama Cloud), or any OpenAI-compatible base URL (Groq, Together, OpenRouter, vLLM…).

Call the gateway from your app

Three env vars are injected when AI is attached: AI_GATEWAY_URL, AI_GATEWAY_KEY, AI_MODEL.

const r = await fetch(process.env.AI_GATEWAY_URL + "/chat", {
  method: "POST",
  headers: { "content-type": "application/json", authorization: "Bearer " + process.env.AI_GATEWAY_KEY },
  body: JSON.stringify({ messages: [{ role: "user", content: "Summarize this." }] }),
})
const { content } = await r.json()

Pass "stream": true for a Server-Sent Events stream. Override the model per-call with "model". The same request shape works no matter which provider is behind the gateway.

Vectors are built in

Your app's database has pgvector — CREATE EXTENSION IF NOT EXISTS vector and store embeddings for RAG. Generate embeddings through the gateway, store them in your DB, done.

Scaffold an AI app

wess init myapp --template ai