Getting Started
DAEMI runs locally. The goal of this guide is to get you from zero to a running session with your first agent.
DAEMI is in private beta. To request access, contact the DAEMI team. You'll receive credentials and setup instructions once approved.
Installation
DAEMI is distributed as an installer. Beta participants receive a download link when approved.
Run the installer and follow the on-screen steps. It will:
- Set up the DAEMI server and all dependencies
- Create the DAEMI config directory
- Copy default agent profiles to get you started
- Configure the Console
Prerequisites
Before you begin, you'll need:
- An LLM provider key — Either an OpenAI API key or an Anthropic API key
- Beta access — See above
Setup Steps
1. Set Your LLM Provider
DAEMI supports two LLM providers. Set your preferred one in the server config:
# For OpenAI (gpt-5.3-codex via Responses API)
export DAEMI_LLM_PROVIDER=openai-codex
export OPENAI_API_KEY=sk-...
# For Anthropic
export DAEMI_LLM_PROVIDER=anthropic
export ANTHROPIC_API_KEY=sk-ant-...
Or set these in the Console under Settings → System.
2. Start the Server
daemi start
You should see:
DAEMI v0.x.x starting...
Loading profiles...
Loaded 3 agent(s): Koda, Aria, Tema
Memory store initialized
Tool graph loaded: 37 nodes
Server running
3. Open the Console
Open the Console in your browser. The Console will connect to your local DAEMI server automatically.
4. Start a Session
- Click New Session in the Console
- Select an agent from the list (e.g., "Koda")
- Start chatting
Your first message kicks off a session. DAEMI will load the agent's profile, initialize its memory context, and begin the conversation.
What to Expect
- First session: The agent starts fresh — no long-term memory yet, but its personality and role are active
- Subsequent sessions: The agent will surface relevant memories from past interactions
- Memory builds over time: The more you use an agent, the richer its LTM becomes