--- title: "Deploy Mem0 (w/ Dashboard)" description: "Self-hosted Mem0 memory layer for AI agents, with dashboard and pgvector" category: "AI/ML" url: https://railway.com/deploy/mem0-w-dashboard --- # Deploy Mem0 (w/ Dashboard) Self-hosted Mem0 memory layer for AI agents, with dashboard and pgvector **[Deploy Mem0 (w/ Dashboard) on Railway](https://railway.com/template/mem0-w-dashboard)** Machine-readable deploy manifest (JSON, validated by TemplateCI): https://railway.com/deploy/mem0-w-dashboard/manifest.json - **Creator:** Auromations - **Category:** AI/ML - **Total deploys:** 1 ## Template content ### mem0 https://raw.githubusercontent.com/mem0ai/mem0/main/docs/logo/favicon.png - **Image:** ghcr.io/hmseeb/mem0-api:latest - **Health check:** /auth/setup-status - **Public domain:** Yes ### pgvector https://cdn.jsdelivr.net/gh/homarr-labs/dashboard-icons/svg/postgres.svg - **Image:** pgvector/pgvector:pg17 ### dashboard https://raw.githubusercontent.com/mem0ai/mem0/main/docs/logo/favicon.png - **Image:** ghcr.io/hmseeb/mem0-dashboard:latest - **Health check:** /api/health - **Public domain:** Yes ## Documentation # Deploy and Host Mem0 on Railway Mem0 is a long term memory layer for AI agents and assistants. It reads a conversation, decides what is worth keeping, and stores it as searchable facts so your agent remembers a user across sessions instead of starting cold every time. This template runs the self hosted mem0 server together with its web dashboard, so memories are something you can browse, search and delete by hand rather than only reach through code. ## About Hosting Mem0 Three services deploy together: the API server, the web dashboard, and Postgres with the vector extension. Memories and their change history sit on persistent volumes, so redeploys do not lose anything. Authentication is on from the first request. The first visit to the dashboard shows a setup form, and the account you create there becomes the only admin. Registration closes the instant it exists, so a public URL is not an open door. From there you issue API keys for your agents and revoke them when a key leaks. The images are built from a pinned mem0 release rather than pulled from Docker Hub, because mem0's own published server image is built for arm64 only and cannot run on Railway's hardware at all. These images also run database migrations at boot and answer on both IPv4 and IPv6, two things Railway needs and the upstream image does not do. ## Common Use Cases - Give a chatbot or support agent memory of a user across separate sessions - Personalise an assistant with stated preferences, constraints and history instead of stuffing everything into the prompt - Keep a private, self hosted memory store when sending conversation history to a managed service is not an option - Inspect and correct what an agent has decided to remember, using the dashboard ## Dependencies for Mem0 Hosting - Postgres with the pgvector extension, included in this template - An OpenAI API key, which you supply at deploy time ### Deployment Dependencies - mem0 source: https://github.com/mem0ai/mem0 - Self hosting docs: https://docs.mem0.ai/open-source/features/rest-api - Images and build definitions: https://github.com/hmseeb/mem0-railway - OpenAI API keys: https://platform.openai.com/api-keys ### Implementation Details **What you need before deploying.** An OpenAI API key. Memory extraction and embedding both call OpenAI and there is no bundled local model, so memory writes fail without one. Everything else is generated for you. Prefer Gemini? After deploying, an admin can switch the LLM and embedder to Gemini (or the LLM alone to Anthropic) with a `POST /configure`; the choice persists across redeploys. **First boot.** Migrations run before the server starts. If Postgres is not ready yet the API retries every two seconds for up to a minute, so a short spell of restarts right after deploying is normal and not a failure. **Setting up your account.** Open the dashboard address and fill in the setup form. That first account is the admin, and registration closes as soon as it is created. Nobody else can claim your instance afterwards. There are no default credentials to rotate, because none are seeded. **Connecting an agent.** Create an API key in the dashboard and send it in an `X-API-Key` header. mem0's own `MemoryClient` package will not work against a self hosted server, because it sends a differently shaped header and checks in with mem0's hosted service on startup. Use plain HTTP: ```bash curl -X POST https://your-api.up.railway.app/memories \ -H "X-API-Key: m0sk_..." \ -H 'Content-Type: application/json' \ -d '{"messages":[{"role":"user","content":"I am allergic to shellfish"}],"user_id":"alice"}' ``` The full endpoint list is on the API's own `/docs` page. **Changing models.** `MEM0_DEFAULT_LLM_MODEL` decides what is worth remembering and `MEM0_DEFAULT_EMBEDDER_MODEL` powers search. Anthropic and Gemini are bundled as alternatives and can be selected through the API's `/configure` endpoint after deploying. **Custom domains.** If you move the dashboard to your own domain, update `DASHBOARD_URL` on the API service to match. It is the only browser origin the API accepts, and a mismatch shows up as a dashboard that loads but where every action fails. **Telemetry.** mem0 sends at most two anonymous events per install, when the first admin is created and when the setup wizard finishes. This template leaves upstream's default in place rather than changing it quietly. Set `MEM0_TELEMETRY` to `false` on the API service to opt out. **Scaling and limits.** Everything runs single instance. The memory change history is a file on the API service's volume, so that service does not scale horizontally without moving it. Postgres is a plain container rather than a managed cluster; for production volume, point the API's Postgres settings at a managed database and keep the vector extension enabled there. ## Why Deploy Mem0 on Railway? Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it. By deploying Mem0 on Railway, you are one step closer to supporting a complete full stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway. Mem0 in particular benefits from it: three services that have to find each other over a private network, two persistent volumes, and generated secrets are all handled by the deploy itself, so the only decision left to you is which OpenAI key to use. ## Similar templates - [Chat Chat](https://railway.com/deploy/-WWW5r) — Chat Chat, your own unified chat and search to AI platform. - [Möbius v0.1 Personal AI OS](https://railway.com/deploy/mobius-v01-personal-ai-os) — Self-hosted AI agent that builds and runs apps for you, with memory. - [stella](https://railway.com/deploy/stella) — Self-host stella with web, API, Postgres, Redis, and object storage. Open this page in a browser: https://railway.com/deploy/mem0-w-dashboard