How a 4-Person Team Tripled Content Output Using AI

When I joined Arloa in 2024, we had four people and a product that needed to do the work of a much bigger team.

Arloa was a GenAI platform that helped parents navigate special education — IEP meetings, school advocacy, understanding a system that wasn't designed to be understood. The audience was non-technical by definition. Parents under real stress, not there to figure out your interface. That context shaped every design decision I made.

My first job was the product. I owned UX strategy and visual design end-to-end, building the design language from scratch. For an audience in high-stakes situations, clarity isn't a nice-to-have. It's the product. Every flow had to work for someone who's exhausted, overwhelmed, and needs an answer fast.

But design alone wasn't going to cut it. A four-person team can't run a content operation manually and build a product at the same time. So I built a system to do it.

The content engine runs on Claude Code. Here's how it works: it pulls from RSS feeds covering special education news, generates plain-language summaries written for our audience, pairs each piece with relevant imagery, and queues it for email distribution — all without anyone touching it manually. I built it in VS Code, iterated through Claude Code, and connected it to our email stack. The whole pipeline runs automatically.

Before: four pieces of content per month, produced by hand. After: 12+ pieces per month, with one person occasionally reviewing output instead of creating it. Three times the volume. Zero added headcount.

I also built the acquisition side — landing pages, onboarding sequences, lifecycle automations. The same system-over-screen thinking. Design the flow, not just the page. Those flows drove 40% quarter-over-quarter user growth for a team with no dedicated growth function.

Claude Code was the tool that made most of this possible. It let me contribute directly to the production codebase without being a full-stack engineer. Iteration cycles that used to take weeks — back-and-forth with an outside developer — compressed to days. Sometimes hours. That compression is what made everything else possible.

Here's what I actually learned: there's a difference between using AI to do tasks faster and using AI to build systems that run without you. The content engine didn't need me to wake up Monday and write three summaries. It ran. That's the shift — from AI as a tool to AI as infrastructure.

A four-person team with the right setup can match the output of an organization ten times its size. Not because AI is magic. Because most of what slows teams down is coordination overhead, manual repetition, and context switching. Design those away and what's left is the actual work.