Deployment Options
Where eMemory can run and how its stacks connect: one stack on one machine, a stack with no internet access, stacks peered into a mesh, a local stack peered with eMemory SaaS, and a multi-site enterprise mesh. Each option links to the operator guide that sets it up.
Each option is the same Docker Compose stack; what differs is where it runs and which other stacks it is peered with. See microservice architecture for what one stack contains, and distributed mesh architecture for peering and exchange.
A Single Stack on One Machine
The default, and isolated out of the box. One stack on a laptop or a server, with its database and its archive folder on that machine; agents and programs on the same machine reach the API service and the console; nothing off the machine can.
Stacks in a Mesh
Run a fleet of agents entirely inside your walls. Each machine hosts its own agents and its own eMemory; they peer over your LAN or across sites, and each holds the lanes it has accepted. No cloud, no internet: a private memory mesh that never leaves the building.
Local Stacks Peered with eMemory SaaS
Stacks hosted locally — on laptops or on your own servers — are peered with a stack on eMemory SaaS.
A Multi-Site Enterprise Mesh
One eMemory per region or site, each serving MCP to its own agents and people, for low latency and data residency. Sites peer directly, and every team reads and writes one governed, audited body of knowledge with no central chokepoint: the memory bus is the mesh itself, not a hub.
Patterns Built on the Mesh
Every pattern below uses the same two choices — an owner shares a policy, a peer accepts it — and the same agent profiles. Combine them freely: one engine can be in several patterns at once.