Vane

Why deploy vane on Render?

Vane is a self-hosted AI answering engine that combines web search with LLM-powered responses. It bundles SearXNG for search, supports multiple LLM providers, and provides source citations with persistent search history. It solves the problem of running a private, customizable AI search assistant without relying on third-party services.

This template pre-configures Vane with bundled SearXNG search, Nginx Basic Auth protection, and a persistent disk for local database and search history—all wired together in a single Docker service. Instead of manually setting up Nginx auth, configuring SearXNG as a separate service, and managing disk mounts, you get a working AI search engine with one click. Render handles the TLS certificate automatically and the Blueprint generates your auth credentials at deploy time, so the UI is protected before you even configure your LLM providers.

Architecture

What you can build

After deploying, you'll have a self-hosted AI search interface at a public HTTPS URL where you can ask questions, get answers with cited web sources, and compare responses across different LLM providers like OpenAI, Anthropic, or your own Ollama instance. The instance runs its own bundled search engine, so there's no external search API to configure, and your search history and uploaded files persist across redeploys on attached storage. You'll finish setup by entering your LLM provider API keys through the web UI after the first deploy.

Key features

• Bundled SearXNG search: SearXNG runs inside the same container, eliminating the need to deploy or configure a separate search service.
• Nginx Basic Auth gate: Nginx sits in front of the Vane app and requires authentication before the setup screen is accessible.
• Persistent disk storage: A 5 GB disk mounted at /home/vane/data preserves the local database, search history, and uploaded files across deploys.
• Multi-provider LLM support: Configure OpenAI, Anthropic, Gemini, Groq, or local Ollama endpoints through Vane's setup screen after deployment.
• Pinned Docker image version: The Dockerfile pins Vane to a specific release tag (v1.12.2) to ensure repeatable deployments.

Use cases

• Solo developer deploys private AI search without managing separate infrastructure
• Security researcher compares answers from multiple LLM providers on sensitive queries
• Small team shares a password-protected research assistant for internal projects
• Technical writer keeps persistent search history and uploads for documentation work

What's included

Next steps

1. Open the *.onrender.com URL and sign in with username 'vane' and the BASIC_AUTH_PASSWORD from your service's Environment page — You should see Vane's setup screen prompting you to configure your first LLM provider
2. Configure an LLM provider (OpenAI, Anthropic, or Ollama) in the setup screen and submit a test search query — You should see search results with cited sources and an AI-generated answer within 15 seconds
3. Test search history persistence by refreshing the page or closing and reopening your browser — You should see your previous search query still listed in the history sidebar

Resources

• Vane official site