TopTools

Problem

Most “tools” sites are either hand-built one-offs that never expand or thin AI slop with no keyword strategy. Scaling a library of useful finance and career tools by hand does not keep up with search demand—or quality bar.

Approach

Designing TopTools as a self-growing product: agents research keywords (full research, volume, keyword difficulty, CPM), decide whether a tool is worth building, implement the tool, open a PR, and run a complete SEO pass once selected. Humans review; the loop is built for continuous shipping.

What was unique

Not a chatbot wrapper and not a static directory. The system owns the full pipeline from opportunity scoring to implementation and distribution—so the site can compound without every page starting from a blank editor.

Outcome

Work in progress / shipping narrative: core agent loop and product framing in flight, with finance and career tools as the first verticals. Built to grow itself once the research → build → PR → SEO path is reliable.

• Agent loop designed around volume, KD, and CPM—not vibes
• Selected tools go from decision → implementation → PR → SEO
• Finance & career verticals chosen for durable search intent
• Human-in-the-loop review so autonomy does not mean unreviewed slop

Process

1. 01

   Define the autonomous loop

   Mapped stages: keyword research, opportunity scoring, build decision, implementation, PR, and SEO—each with a clear success check before the next stage.

2. 02

   Pick verticals with intent

   Finance and career tools first—high intent, repeatable formats, and room for a library that compounds in search.

3. 03

   Ship, then expand

   Prove one tool end-to-end through the agent pipeline, then widen the catalog without rewriting the growth system each time.

Stack

• TypeScript
• Next.js
• AI agents
• SEO
• Node.js
• PostgreSQL