Typed SDKs
First-class TypeScript and Python with helpful types, retries, and streaming ingestion built in.
{ developer platform }
Ship memory in minutes,
not months.
A clean API, typed SDKs, first-class MCP support, and deployment options that fit your stack. Start with five lines and scale to millions of memories.
Get your API key$ npx @matiai/cli init
MatiAIProject configuration
- 01FrameworkNext.js
- 02Memory scopePer user
- 03RegionUS East
$ matiai dev
Memory layer readyListening on localhost:8787
42msTypeScriptPythoncURL
import { MatiAI } from "@matiai/sdk";const memory = new MatiAI({ apiKey: "ma_live_..." });await memory.remember({userId: "maya_123", content: "I lead product at Northstar and prefer async updates."});const context = await memory.recall({userId: "maya_123", query: "How should I share this update?"});200 OKContext retrieved in 67ms
QUICKSTART
From zero to recall in five steps.
Most teams have durable context running in production the same day they install the SDK.
01
Install
Add the SDK to an existing Next.js, Node, or Python app in under a minute.
02
Authenticate
Drop in an API key, or exchange an OAuth token for per-user memory scopes.
03
Remember
Send a message, document, or event — extraction happens automatically.
04
Recall
Fetch ranked context for any prompt with a single call.
05
Deploy
Ship to our cloud, your VPC, or self-hosted — the API is identical everywhere.
SDK EXAMPLES
The same idea, in three flavors.
Typed SDKs for TypeScript and Python, plus plain HTTP for everything else. No vendor lock-in, no hidden runtime.
from matiai import MatiAI memory = MatiAI(api_key="ma_live_...") memory.remember( user_id="maya_123", content="Prefers async updates.", ) ctx = memory.recall( user_id="maya_123", query="How to share this?", )curl https://api.matiai.in/v1/memories {\ -H "Authorization: Bearer ma_live_..." \ -H "Content-Type: application/json" \ -d {"userId":"maya_123",\"content":"Prefers async updates."}await memory.namespaces({create: { name: "prod-agent" }});await memory.remember({userId: "maya_123", namespace: "prod-agent", content: "Use Postgres for persistence."});EVERYTHING YOU NEED
From first request to full scale.
The platform primitives you would otherwise build yourself — typed, observable, and production-ready on day one.
Realtime webhooks
React to extraction, updates, deletion, and sync events the instant they happen.
Namespaces
Isolate users, workspaces, agents, and environments cleanly with one API.
Deploy anywhere
Use our cloud, your VPC, or a self-hosted installation — identical APIs.
Per-user auth
Exchange OAuth tokens for scoped, per-user memory without exposing your API key.
Batch ingestion
Import existing conversations and documents with jobs, progress, and deduplication.
Regional routing
Pin memory to US, EU, or APAC regions per workspace for residency and latency.
MCP server
Expose memory to Claude, Cursor, and any MCP-compatible client in one command.
API REFERENCE
The core endpoints.
A small, stable surface area. Everything you can do in the dashboard is available over the API.
| Method | Path | Description | SDK |
|---|---|---|---|
| POST | /v1/memories | Ingest a message, document, or event and extract structured memories. | remember |
| POST | /v1/recall | Retrieve ranked context for a user, namespace, and query. | recall |
| GET | /v1/users/:id/memories | List a user's memories with filters and pagination. | list |
| DELETE | /v1/memories/:id | Delete a single memory or a filtered batch. | forget |
| POST | /v1/namespaces | Create an isolated scope for users, agents, or environments. | scopes |
| POST | /v1/webhooks | Subscribe to extraction, update, deletion, and sync events. | events |
{ Start building }
Give your AI a memory it can trust.
Free to start. Open by design. Ready for production.
Get MatiAI