Skip to main content
Time commitment: 5 minutes Outcome: Complete a real data analysis with SignalPilot’s AI agent

​
Step 1: Install SignalPilot

Prerequisites: macOS, Linux, or Windows (WSL) • Internet connection Install SignalPilot: 
uvx signalpilot
Install uv first (takes seconds):
curl -LsSf https://astral.sh/uv/install.sh | sh
Then run SignalPilot:
uvx signalpilot
See uv installation guide for other methods (Homebrew, Windows, etc.)
What happens: 
✓ Creating workspace at ~/SignalPilotHome
✓ Installing isolated Python 3.12 + Jupyter Lab + SignalPilot extension
✓ Installing data packages (pandas, numpy, matplotlib, plotly...)
⚡ Setup complete in ~2 minutes

→ Launching Jupyter Lab at http://localhost:8888
Option 1: Using uvx (recommended)
uvx signalpilot home
Option 2: Manual activation
cd ~/SignalPilotHome
source .venv/bin/activate && jupyter lab
See the Installation Guide for:
  • Manual installation with pip, conda, or uv
  • Working in different modes (default, home, project)
  • Auto-update configuration
  • Troubleshooting

​
Step 2: Create Your First Notebook

1

Create a new notebook

Your browser opened Jupyter Lab at http://localhost:8888You’ll see the SignalPilot panel in the left and right sidebar.Click File → New → Notebook and select Python 3Name it revenue-analysis.ipynb
Always use the default Python 3 kernel. SignalPilot uses this kernel and may not work with other kernels due to missing system variables.
2

Load your data

Option 1: Use local files (recommended)Type @ in the SignalPilot chat to mention any CSV, Excel, or data file:
  • Home workspace: ~/SignalPilotHome/data/
  • Project folder: ./data/ in current working project directory
Use the File Scanner in the left sidebar to add more folders to index.
Option 2: Generate sample dataIf you don’t have data files, create sample data in the first cell:
import pandas as pd
import numpy as np

# Create sample revenue data
dates = pd.date_range('2025-01-01', '2025-12-31', freq='D')
revenue = pd.DataFrame({
    'date': dates,
    'revenue': np.random.randint(1000, 5000, len(dates)) +
               (np.sin(np.arange(len(dates)) / 7) * 500).astype(int)
})

revenue.head()
Expected: DataFrame loaded from file or displayed from generated data
3

Ask SignalPilot to analyze

Open the SignalPilot chat panel (left sidebar) and ask:
Analyze this revenue data. Show me:
1. Total revenue for the year
2. Average daily revenue
3. A time-series plot showing trends
4. Identify any anomalies or patterns
Expected: SignalPilot creates multiple cells with:
  • Aggregation queries
  • Visualization code
  • Statistical analysis
  • Interpretation of results
4

Iterate on the analysis

Ask a follow-up question:
What's the week-over-week revenue growth rate? Show me a chart.
Expected: SignalPilot adds new cells calculating WoW growth and visualizing it

​
What You Just Learned

SignalPilot automatically understood the revenue dataframe in your kernel and used it without you having to copy-paste or describe it.
The agent broke down your request into multiple steps: aggregation, visualization, and interpretation.
You asked a follow-up question, and SignalPilot used context from the previous analysis to build on it.
SignalPilot wrote executable pandas and matplotlib code directly into notebook cells.

​
Next Steps

Connect Real Data

Connect to your data warehouse

Learn Key Concepts

Understand agents, modes, and context

User Funnel Analysis

Analyze user drop-off with Amplitude and BigQuery

dbt Integration

Connect to your dbt project for lineage

​
Troubleshooting

Solution: Ensure you’ve executed the cell that creates the dataframe. SignalPilot reads from the active kernel state.
Solution: Check that you’re in Agent Mode (not Ask Mode). You can switch modes in the SignalPilot panel.
Solution: Ensure matplotlib is installed: uv add matplotlib