Architecture
How Autonomis works behind the scenes — explained simply.
How Your Agents Run
Think of each agent as a dedicated assistant that lives in its own private room. When you create an agent on Autonomis, we spin up an isolated container — a small, self-contained computing environment — just for that agent. No other user can access it, and your agents never share resources with anyone else.
Your agents run 24/7. They are always on, always ready to respond to your messages, execute tasks, and collaborate with your other agents.
The Building Blocks
Autonomis is made up of a few key components that work together to power your agents:
- Your agents — AI assistants running around the clock in their own containers. Each one has its own personality, knowledge, and skills that you control through templates.
- A database — Stores your settings, agent configurations, usage data, and long-term memories. Your data is always backed up and protected.
- Encrypted storage — Your API keys, passwords, and other secrets are encrypted using military-grade encryption (AES-256) before being stored. They are only decrypted inside your agent's container and are never visible to anyone — including us.
- The dashboard — Your web-based control center where you manage agents, chat, configure templates, add secrets, install skills, and monitor usage. See the Dashboard guide for details.
What Each Agent Gets
Every agent you create receives its own dedicated set of resources. Nothing is shared between agents — each one operates independently:
- Its own container — A private computing environment, like a personal computer just for that agent.
- 200 MHz CPU and 1.5 GB memory — Guaranteed resources so your agent always has the power it needs to run smoothly.
- Its own workspace — A private file storage area where the agent keeps documents, notes, and working files.
- A communication channel — When you have multiple agents, each one can send and receive messages to collaborate with the others securely.
Vector Memory
Your agents have long-term memory powered by AI embeddings — a technology that lets them understand and recall information by meaning, not just exact words.
Think of it like this: instead of searching through notes for a specific keyword, your agent can remember "that conversation about the marketing budget from last week" even if those exact words were never used.
- Automatic indexing — Agents automatically organize their notes and interactions so they can find them later.
- Semantic recall — Agents can search their memory by meaning, surfacing relevant information even if the wording is different.
- Persistent — Memories survive restarts and are stored safely in the cloud, so nothing is lost.
Multi-Agent Communication
When you have multiple agents, they can work together as a team. Agents communicate through a secure messaging system — like an internal chat room that only your agents can access.
For example, you might have a "manager" agent that delegates research to a "researcher" agent and writing to a "writer" agent. The manager coordinates the work, and each specialist focuses on what it does best. Learn more in the Agent Configuration guide.
Security
Your data is encrypted, isolated, and only accessible by you. Every connection between components is encrypted, and each user's data is completely separated from every other user's data. Since you bring your own AI provider key, your conversations go directly between your agent and the AI provider — we never see them.
For a detailed breakdown, see the Security page.
Auto-Scaling
You do not need to worry about capacity. Autonomis automatically adds and removes servers based on demand. When more agents are running across the platform, we scale up. When things quiet down, we scale back. This happens behind the scenes — your agents are always ready.