Skip to content

Programmable memory for agents, apps, and workflows.

Integrate memory into agents, apps, and workflows using an MCP-compatible interface and a clean API surface. Bring your own LLM keys; we run the memory.

MCP-compatible interface

MCP-Compatible Interface

Protocol

The memory layer is internally MCP-compatible, with a limited public MCP interface for read, write, and retrieval. Advanced orchestration and behavioural enforcement remain internal.

MCP read / write

https://mcp.octamem.com

Public endpoints available

Spec · Model Context Protocol

§ 01API surface

Clean, consistent endpoints.

The same memory primitives across every interface. Pick the surface that matches your runtime — the contract is identical between REST, the Python and JavaScript SDKs, and MCP.

  • Authentication

    Bearer · API keys · scopes

    OCTA_API_KEY in the Authorization header.

  • Memory · add

    POST /v1/memory/add

    Capture a record with previous_context and type.

  • Memory · search

    POST /v1/memory/search

    Retrieve typed records, ranked and source-linked.

  • Usage & account

    GET /v1/account · /v1/usage

    Plan, storage used, rate budgets, billing period.

  • Python SDK

    pip install octamem

    Typed, async-first. Works with any async runtime.

  • JavaScript SDK

    npm install @octamem/octamem-js

    TypeScript-first. Edge-runtime compatible.

§ 02Quick start

Quick start examples.

Add a memory, search it, inspect a record. Same shape across Python, JavaScript, and the REST API.

add_memory.py · python
from octamem import OctaMem

client = OctaMem(api_key="YOUR_API_KEY")

client.add(
    content="Launch date set for April 15. Beta opens March 20.",
    previous_context="Q1 product planning meeting",
)
§ 03Use cases

Built for the next generation of AI apps.

  • Chatbots.

    01

    Give chatbots long-term memory across sessions and model switches.

  • AI Agents.

    02

    Agents retain behavioural context and preferences across tasks.

  • Copilots.

    03

    Developer copilots that remember project structure and patterns.

  • RAG Pipelines.

    04

    Augment retrieval with structured user memory for personalised results.

§ 04Ready to integrate

Get your API key. Start building.

Free tier includes 2 GB of memory and full SDK access. Upgrade to team storage, audit log, or VPC deployment when you need them. The code stays the same.