Ticket triage

This example uses DSPy with a local deterministic LM to classify a support ticket, assign a team, and draft a short reply.

Requirements

• Python 3.10+

Steps for running this example in the shell

1. Install dependencies:

python -m venv .venv

# macOS/Linux:
source .venv/bin/activate

# Windows Command Prompt:
.venv\Scripts\activate.bat

# Windows PowerShell:
.venv\Scripts\Activate.ps1

pip install -r requirements.txt

2. Run the script:

python ticket_triage.py

What it demonstrates

• A DSPy signature for structured ticket triage
• A local LM stub that returns repeatable JSON
• A dspy.Predict module that parses the JSON into typed outputs

Output

Ticket 1: I was charged twice for my subscription and need a refund.
Priority: high
Team: billing
Reply: Thanks for the details. I am routing this to billing now and will help get it resolved.

Ticket 2: I cannot log in after resetting my password.
Priority: high
Team: support
Reply: Thanks. I am routing this to support so we can help restore your access quickly.

Ticket 3: The dashboard export button is broken and shows an error.
Priority: medium
Team: engineering
Reply: Thanks for reporting this. I am sending it to engineering for a closer look.