Getting started
Sign in, add a model key, create an API key, send your first request.
Memtro sits between the tools your team uses and the models you pay for. This page takes you from a fresh sign-in to a working API key in about five minutes.
1. Sign in
Go to platform.memtro.com and create an account with email and password, or sign in with Google. Password accounts set up an authenticator app (Google Authenticator, Authy, 1Password…) straight after sign-up, and sign-in asks for a six-digit code after the password from then on. Keep the backup codes it gives you somewhere safe. If you were invited to an organisation, the invitation is waiting on the Organisation page after you sign in.
2. Create or join an organisation
Everything in Memtro lives inside an organisation: model keys, connections, memory and jobs are shared with it, with personal items kept to you. Right after sign-up you either accept a pending invitation (joining uses that organisation's subscription) or create your own organisation and start its 7-day free trial: a card is taken, nothing is charged until the trial ends, then it is $50 a month per organisation. Owners and admins manage billing and shared resources; members use them. You can belong to several organisations and switch between them from the sidebar; each has its own subscription. See pricing.
3. Add a model provider
On Dashboard → Providers pick a provider card and add a key:
| Provider | What to paste | Notes |
|---|---|---|
| Anthropic | Console API key (sk-ant-api03-…) | Needed for client-declared tools and the fastest path to Claude. |
| Claude Code | The token printed by claude setup-token | Runs requests through the Claude Code CLI on your subscription. See the Claude Code guide. |
| OpenAI | API key (sk-…) | Also provides text-embedding-3-small for memory retrieval. |
| Google Gemini | API key (AIza…) | Also provides gemini-embedding for memory retrieval. |
| OpenAI-compatible | Label, base URL and key | Groq, Together, OpenRouter, Ollama, vLLM or any /v1 chat endpoint. Addressed as compat:<label>/<model>. |
Memory retrieval needs embeddings, which come from an OpenAI or Google key. Add one of those even if you plan to chat through Anthropic or Claude Code.
Personal keys take precedence over shared keys for the same provider.
4. Create an API key
Dashboard → API keys issues keys that start with sk-memtro-. A key acts as you, inside the organisation it was created in. Store it like any other secret; it is shown once.
5. Send a request
curl https://platform.memtro.com/v1/chat/completions \
-H "Authorization: Bearer sk-memtro-..." \
-H "Content-Type: application/json" \
-d '{"model":"auto","messages":[{"role":"user","content":"Who owns the billing migration?"}]}'
auto lets Memtro pick a model by how hard the question is. memtro uses the organisation's default model. anthropic/claude-sonnet-5, openai/gpt-5.4 or claude-code/opus name one directly. See the API reference.
6. Try the Chat page
Dashboard → Chat talks to the same endpoint with your session, defaults to auto in agent mode, and keeps every conversation so you can come back to it. Tell it something ("Remember that Priya owns the billing migration, launching 14 November"), then ask about it in a new chat.
Next steps
- Connect your systems so the model can look things up for real.
- Point your tools at Memtro: Cursor, Open WebUI, OpenCode, Claude Code, the OpenAI SDKs.
- Schedule a job for the recurring work.