morphic

Why we included this project

Morphic takes a natural-language question and returns a synthesized, cited answer rather than a bare list of links, which makes it a reasonable choice for teams that want an AI search experience without routing their queries through a commercial product. Two search modes cover the common cases: quick for fast results, adaptive for deeper, multi-step questions. The generative UI is the standout part: answers stream in as live components built from a JSON spec, so source-credited images and headings render inline instead of as plain markdown. You can run it against OpenAI, Anthropic, Google, Ollama, or any OpenAI-compatible endpoint, and pair it with Tavily, SearXNG, or Brave as the search backend. A bundled Docker Compose setup pulls in PostgreSQL, Redis, and SearXNG automatically, so standing up the full stack locally is straightforward, and it also deploys to Vercel when you want a hosted instance.