SuprfloThe memory layer for AI agents

Extract facts from conversations, store them with validity windows, and recall exactly what your agent needs — through one REST API, every SDK, and 13 framework integrations.

Get started

REST API

GitHub

Python SDK

pip install suprflo — MemoryClient with add, search, get, update, delete, and framework adapters behind extras.

Python SDK

JavaScript / TypeScript SDK

npm install @suprflo/sdk — the same memory API for Node and the browser.

JS / TS SDK

LangChain

Drop-in memory, a retriever, and agent tools for LangChain chains and agents.

LangChain

LangGraph

A BaseStore-backed memory for stateful LangGraph agents.

LangGraph

LlamaIndex

Memory and a retriever that returns scored nodes for LlamaIndex.

LlamaIndex

CrewAI

External storage and tools so a crew can save and recall memories.

CrewAI

AutoGen

The autogen_core Memory protocol, backed by Suprflo.

AutoGen

Voice — LiveKit · Pipecat · ElevenLabs

Recall before the LLM turn, persist the transcript after — for real-time voice agents.

Voice integrations

Vercel AI SDK & Mastra

Memory tools and prompt recall for generateText / streamText and Mastra agents.

TypeScript frameworks

Pluggable backends

17 LLMs, 11 embedders, and 9 vector stores — swap the model or database by environment variable.

Backends