Note: Setup has two parts — (1) connecting it to the Lucidic Dashboard (below), and (2) installing LucidicAI in your code.

​
Getting Started in the Dashboard

Before you can track your agent’s sessions in code, you’ll need to set up your environment in the Lucidic web dashboard.

​
1. Create an Account or Sign In

Go to dashboard.lucidic.ai and sign in:
  • Use Google login or
  • Create an account with your email + password

​
2. Create a Project

Projects are high-level containers for your work — usually one per app, workflow, or team.
  • Click “Create Project”
  • Name it something meaningful (e.g. wikipedia-agent, finetuning-dashboard, etc.)

​
3. Create an Agent

Inside your project:
  • Click ”+ Agent”
  • Provide a name and choose an icon
Agents represent individual autonomous systems (e.g. a browser agent, a research assistant, a planner).

​
4. Generate an API Key

You’ll need an API key to authenticate your SDK.
  • Go to the “API Keys” icon in the bottom left corner in the side bar
  • Click “Create API Key”
  • Copy the key somewhere safe - you won’t be able to see it again!
api key

​
5. Plug Into Python

Once you’ve got:
  • API Key
  • Agent ID
You’re ready to run sessions from code. Paste them into: 
lai.init(
    api_key="your-api-key",
    agent_id="your-agent-id"
)
Or set them as environment variables: 
export LUCIDIC_API_KEY="your-api-key"
export LUCIDIC_AGENT_ID="your-agent-id"

​
6. (Optional) Create a Mass Simulation

If you want to run many sessions as part of a single evaluation:
  • Go to the “Mass Sims” tab inside your Agent
  • Click “New Mass Simulation” on the top right of the table
  • Copy the Mass Sim ID into your SDK: 
# assuming environment variables are set
lai.init(
    mass_sim_id="abc123"
)
This will link that run to your high-level simulation — perfect for testing prompt changes, model swaps, or behavior analysis at scale. To learn more about mass simulations, see Mass Simulations. To learn more about creation of mass simulations, see Creating a Mass Simulation.

​
You’re Ready!

Now that your dashboard is set up, head to Quick Start to run your first session with just 2 lines of code.