← Back to multiprovider.cloud

How It Works

The technical overview, in plain English.

The short version: a lightweight memory router sits between you and your AI providers. It stores your context, routes each query to the best available model, and returns the result — without touching your data beyond what you explicitly share. Your memory lives in a graph on a disk you own.

The problem it solves

Every AI model forgets you the moment the conversation ends. The big providers will happily sell you their memory — but it's their memory of you, in their silo, that you rent back month after month, and that you lose the day you switch models. Your reasoning history shouldn't be a hostage.

The router

1. You own the memory

Your context, preferences, and reasoning history are stored in a graph — on infrastructure you control. Self-host it on your own hardware, or use our hosted option. Either way, it's yours: portable, inspectable, and not locked to any one vendor.

2. It routes to the right model

When you ask something, the router picks the best available model for that task and hands it the context it needs — no more, no less. Different jobs want different models; you're never locked into one vendor's strengths and weaknesses.

3. It returns the result — and remembers

The answer comes back, and the exchange is written to your memory graph, so the next conversation builds on the last. Over time your AI relationship compounds instead of resetting.

What it does NOT do

Free, and open

The code is open-source. Self-hosting is free (you pay only for your own infrastructure, and we give you an honest estimate). The hosted option is free too — a small model, one-click installer, no credit card. See what it costs.


Get Started Free Know All the Facts


Made with care by The HPL Company · Privacy · Terms