Create a reminder for a user. Accepts natural language ("tomorrow 9am call Andrei") or a strict ISO-8601 datetime + body. Returns the parsed schedule so you can confirm before notifying.
Arvis API for developers
A small, honest REST surface for the things Arvis is best at: reminders, messages, files, usage. EU-hosted, bearer-auth, no surprises.
Closed beta · opening Q3 2026What you can do today
The honest versionRight now Arvis is conversation-first. Everything happens in Telegram. The API exists in working form internally (we use it ourselves) but is not yet open to outside customers. If you are on the Apex tier and want early-access keys before the public Q3 opening, drop us a note from inside the bot with /api_request — we are onboarding teams in small batches.
Planned endpoints
v1 surface area · subject to changeList a user's active reminders. Cursor-paginated (50 / page), filterable by status & due_before. Includes the original phrasing — useful for analytics dashboards.
Cancel a reminder. Idempotent — deleting an already-cancelled reminder returns 204, never 404. The user is notified in-bot that the reminder was cancelled by an external integration.
Send a message through Arvis to one of your users. Goes out as a normal bot message — respects their language, quiet hours, and per-user mute. Great for proactive integrations (deploy notifications, on-call paging, CRM events).
Per-user and per-key token consumption, broken down by model family. Useful for billing pass-through and for spotting accidental loops before they get expensive.
Upload a PDF, DOCX, or plain-text file and ingest it into a user's RAG index. After this, the user can ask questions against the document from inside the chat. Returns the resulting chunk count and an ingest receipt.
Authentication
Bearer token via Apex dashboardEvery Apex subscription includes one API key, generated from the in-bot /api_keys screen. Rotate keys at any time — old keys keep working for 24h so deploys never break mid-flight.
Rate limits
Apex tier baseline · higher on requestHitting a limit returns 429 with a Retry-After header. If you need more headroom for a real production use case, tell us — we'll lift the cap.
What you will not get from this API
Honest scopeThis is a focused API for the specific things Arvis does well in chat: reminders, messages, files, usage. It is not a general-purpose LLM API — for that, use Anthropic, OpenAI, or Mistral directly. The Arvis API exists so you can wire Arvis-the-product into your stack, not so you can rent inference.
Get early-access keys
Closed beta opens to Apex customers first. The fastest way in is to upgrade and send /api_request inside the bot — we review every request manually so we can size capacity correctly.