Supermemory with Cloudflare Zero Trust

Self-hosted Supermemory memory engine—ingestion, hybrid semantic search, and full Memory API—protected behind Cloudflare Zero Trust Access.

Why Deploy

Supermemory brings persistent, semantic memory to your applications. Use it to:

Store and retrieve documents with hybrid search (keyword + embedding)
Extract structured insights from unstructured data
Build AI agents with long-horizon memory and context
Keep sensitive data on your own infrastructure with full privacy

Cloudflare Zero Trust ensures only authenticated users can access your memory engine.

Deploy and Host

This template deploys a Supermemory server on Railway with a 10GB persistent volume. Choose your LLM provider during deployment:

OpenAI (gpt-4o recommended) for production workloads
Anthropic Claude for privacy-first setups
Ollama for fully local/offline operation
Default embeddings (no API keys) for cost-free operation

Cloudflare Zero Trust credentials protect the public endpoint. Set CF_ACCESS_CLIENT_ID, CF_ACCESS_CLIENT_SECRET, and CF_ACCOUNT_ID from your Cloudflare dashboard.

About Hosting

Railway provides the compute, networking, and persistent storage:

Compute: CPU/RAM scales automatically with your plan
Volume: 10GB persistent storage for the embedded graph database
Networking: Public domain with private database connections and Cloudflare Access overlay
Monitoring: Health checks on the /health endpoint

Common Use Cases

AI-powered customer support: Store interactions, retrieve relevant context for agents
Research assistant: Ingest PDFs/documents, semantic search across your library
Knowledge management: Centralized memory for team workflows
Secure data analysis: Keep sensitive datasets behind Zero Trust

Dependencies for

Supermemory is self-contained—no external databases required. The embedded graph engine persists to the Railway volume. Optional dependencies:

LLM API: OpenAI, Anthropic, or Ollama (for intelligence; works without)
Cloudflare Access: For Zero Trust protection (or skip for open access)

API Reference

Supermemory exposes the full Memory API:

POST /v3/documents — Add a document
POST /v4/search — Hybrid semantic search
GET /v4/profile — Memory profile
GET /health — Health check

See Supermemory docs for SDK examples (JavaScript, Python, Go, Rust).