AI Case Studies

Introduction

Real production case studies — read these to understand how leading AI products were actually built and what trade-offs they made.

Understanding the topic

Core concepts to understand:

• Klarna — replaced 700 support agents with one AI assistant.
• Cursor — built the dominant AI code editor on top of multiple models.
• Perplexity — productionised research-agent UX.
• Notion AI — RAG over private workspace data at huge scale.
• Duolingo Max — GPT-4 roleplay tutor that became a top retention feature.
• Bloomberg GPT — full pre-train on financial data.
• Harvey AI — legal vertical agent with deep domain expertise.

Syntax reference

Visual workflow / architecture:

Case study reading checklist:
  - Problem solved
  - Architecture chosen
  - Models used
  - Evals + metrics
  - Cost & scaling
  - Lessons learned

Real-world use

Most case studies are documented in engineering blogs, conference talks (e.g. AI Engineer Summit) and post-mortems.

Best practices

• Read one case study per week.
• Take notes on architecture + trade-offs.

Hands-on exercise

Interview preparation — practice these questions:

• Q1. Pick a case study and explain its architecture.
• Q2. What would you do differently?
• Q3. What evals would you add?