The Two-Agent System I Use to Build SaaS Without Over engineering

AI is great at writing code quickly. The problem is that it often creates too much of it.

A simple feature can turn into extra interfaces, dependencies, services, and abstractions for problems you may never have. AI tends to prepare for every possible future instead of helping you ship the next useful version.

I use two AI agents to keep that under control.

Step 1: The Architect

I start with an Agency Agent, usually focused on software architecture, SEO, or marketing.

I give it the problem, and it creates a detailed plan. It looks at the database, routes, UI, dependencies, and edge cases.

This gives me a strong starting point, but the plan is usually too large.

The Architect designs the perfect system. As a solo founder, I usually need a practical one.

Step 2: Ponytail

Before I execute the plan, I send it to Ponytail.

Ponytail is a cynical senior developer persona with one rule:

The best code is the code you never write.

Ponytail asks:

• Do we need this now?
• Can a browser feature replace a new package?
• Can three interfaces become one function?
• Are we solving a real problem or an imaginary one?

The Architect may suggest a state management library for a basic form. Ponytail may replace it with standard HTML form data.

The Architect expands the plan. Ponytail cuts it down.

Why this works

The Architect helps me see the whole problem. Ponytail stops me from building more than I need.

The process is simple:

1. Create a complete plan.
2. Challenge every part of it.
3. Remove speculative work.
4. Build the smallest version that solves the problem.

This works for code, landing pages, marketing copy, and bug fixes.

AI can generate a lot of work quickly. It still needs an editor who knows what to remove.

Curious what this process produces? Check out DataCurlew.in.