AI SaaS Applications

Introduction

An AI SaaS is a subscription product where the core value is an AI workflow. Examples: Cursor (AI IDE), Perplexity (AI search), Jasper (AI copywriting), Harvey (AI legal). Building one combines product, infra, and prompt engineering.

Beginner analogy: Like building any SaaS, but the killer feature is the LLM under the hood.

Understanding the topic

Core concepts to understand:

• Pick a clear vertical with painful workflows.
• Wrap LLM with proprietary data, integrations, and UX.
• Pricing: usage-based, seat-based, or hybrid.
• Build evals + feedback loops from day one.

Syntax reference

Visual workflow / architecture:

Vertical pain point
       │
       ▼
LLM + your proprietary data + integrations
       │
       ▼
Polished UX + onboarding
       │
       ▼
   Pricing model
       │
       ▼
   Land + expand

Real-world use

Cursor ($100M+ ARR in <2 years), Perplexity, Glean, Decagon, Sierra — the AI SaaS gold rush is real.

Best practices

• Own the data layer — pure-LLM wrappers get commoditised.
• Track per-customer cost vs revenue closely.
• Build evals before scaling.

Hands-on exercise

Interview preparation — practice these questions:

• Q1. What separates a good AI SaaS from a 'GPT wrapper'?
• Q2. How do you defend a moat?
• Q3. What are common pricing models?