What it is

Morphic is an open-source, AI-powered search engine and answer engine with a generative user interface. It lives in the TypeScript, Next.js, React, and Vercel AI SDK ecosystem, and it is distributed under the Apache-2.0 license. The project combines model selection, search providers, chat history, file upload, and shareable result URLs in one web application.

It addresses the problem of AI search results that appear only as plain text or markdown, which can make sources, media, and structured answers hard to scan. Morphic renders answers as rich inline components from a streamed JSON specification, while grounding responses with cited search results. It also supports local and hosted model providers, including Ollama and OpenAI-compatible services, so users can choose the model stack that fits their environment.

Key capabilities

• Morphic performs AI-powered search and returns grounded, cited answers instead of unverified text alone.
• Its generative UI can render source-credited images, grids, and headings inline from a streamed JSON specification.
• The interface includes Quick and Adaptive search modes for different answer-generation workflows.
• A model selector detects and supports OpenAI, Anthropic, Google, Ollama, Vercel AI Gateway, and OpenAI-compatible providers.
• Search can use Tavily, SearXNG, Brave, or Exa, and Docker Compose can provide SearXNG without a separate search API key.
• Chat history is stored in PostgreSQL, results can be shared through unique URLs, and file upload is available.
• Authentication is available through Supabase Auth, while guest mode allows anonymous usage.

Who uses it and how

• Self-hosters run Morphic with Docker Compose, which starts PostgreSQL, Redis, SearXNG, and the application together.
• Developers deploy the app to Vercel with environment variables for an AI provider and a search provider such as Tavily.
• Users search with a selected model and provider, then review answers that include citations and rendered components.
• People can use shared URLs to distribute search results, and authenticated users can retain chat history.
• Individuals can use guest mode for anonymous searches or Supabase Auth for account-based access.

Getting started

Typical installation uses Docker: pull ghcr.io/miurla/morphic:latest, copy .env.local.example to .env.local, set at least one AI provider API key, then run docker compose up -d and open http://localhost:3000. For local development, clone the repository, run bun install, configure .env.local, and start the app with bun dev; the README also lists a Vercel deployment button.

When to use it — and when not to

Use Morphic when you want a self-hostable, open-source AI search front end that supports multiple model providers, multiple search providers, and rich cited answers. Avoid it when you need a fully managed service, because a self-hosted deployment requires operating PostgreSQL, Redis, SearXNG, and provider credentials, and the provided facts show 56 open issues and 0 listed contributors, which means maintenance status should be checked before production use.

docker pull ghcr.io/miurla/morphic:latest

git clone https://github.com/miurla/morphic.git
cd morphic
cp .env.local.example .env.local

OPENAI_API_KEY=your_openai_key

docker compose up -d

git clone https://github.com/miurla/morphic.git
cd morphic
bun install

cp .env.local.example .env.local

OPENAI_API_KEY=your_openai_key
TAVILY_API_KEY=your_tavily_key

bun dev