Skip to main content

Solving NYT Connections with DsPy

In this tutorial, you'll build an AI system that solves NYT Connections puzzles using DsPy for structured reasoning and Dagster for pipeline orchestration. DsPy is a framework for building modular AI software that can be compiled into effective prompts and optimized automatically.

This example demonstrates how to:

  • Load and process Connections puzzle data
  • Build a DsPy solver module with game logic
  • Optimize the solver using MIPROv2 automatic optimization
  • Evaluate puzzle-solving performance with custom metrics
  • Deploy and monitor AI systems in production

Prerequisites​

To follow the steps in this guide, you'll need:

  • Basic Python knowledge
  • Python 3.9+ installed on your system. Refer to the Installation guide for information.
  • Understanding of language models and AI pipelines
  • A Google Gemini API key (get one at Google AI Studio)
  • Familiarity with puzzle-solving and game logic
  • Basic knowledge of the NYT Connections puzzle format

Step 1: Set up your Dagster environment​

First, set up a new Dagster project with DsPy dependencies.

  1. Clone the Dagster repo and navigate to the project:

    cd examples/docs_projects/project_dspy

  2. Install the required dependencies with uv:

    uv sync

  3. Activate the virtual environment:

    source .venv/bin/activate

  4. Set up your environment variables by copying the template:

    cp .env.example .env

    Then populate the required fields:

    • GEMINI_API_KEY: Your Google Gemini API key for language model access
    • GEMINI_MODEL: The Gemini model to use (default: gemini-2.0-flash-exp)
    • DAGSTER_CLOUD_API_TOKEN: (Optional) For Dagster Cloud deployment

Step 2: Launch the Dagster webserver​

To make sure Dagster and its dependencies were installed correctly, navigate to the project root directory and start the Dagster webserver:

dg dev

Navigate to http://127.0.0.1:3000 to view the Dagster UI.

Architecture overview​

The DsPy pipeline consists of several key components:

Next steps​