📋 Table of Contents

• Workshop Prep
• Data Analysis Fundamentals
• Agent Build-Along
• Debrief

Workshop Prep

• Sign up for Google Opal and link it to your Google Account. If you have time, click around and play with the demo agents in their Gallery section. Try importing a file from your Google Drive. We’ll do a full walk-through together.
  • Why Opal? 1. It’s a sneak peek into Google’s future products lab. 2. It works easily with code execution. 3. It’s currently free, so you don’t have to pay additional subscription or for credits used. And 4: it uses a similar mental model as other popular platforms such as Lindy and n8n, so your learning will transfer.
• Have a dataset that you’re excited to dig into? Get it ready by importing it as a Google Sheet, and/or have it saved as a .CSV file. For the purposes of today’s workshop - less is more. Let’s start with 20 clean rows of a dataset that’s relevant to you, vs 10,000 rows of the original.

Data Analysis Fundamentals

LLMs are trained to interpret natural language

• "Every LLM was trained on billions of words, learning to predict what comes next in a sentence” - source

  

  

LLMs struggle with math, data, and large datasets

• "A 50-column table isn't a story, it's a multidimensional relationship map where correlations matter more than sequence”

  

Techniques are being developed to make LLMs better at working with data

• Example 1: Program-Aided Language Models (PAL) that generate Python or SQL queries

  

• Example 2: Systems that leverage Symbolic AI

  

Performance is improving but often "spiky”

• DSBench eval was built specifically for data science