# LawnCare.center

A Python pipeline turns keywords and YouTube videos into QA-scored, affiliate-linked articles. A Next.js site renders them.

URL: https://jaketaylor.design/work/lawncare-center

- Year: Nov 2025—Present
- Development: Claude Code
- Tools: Next.js 16, React 19, Tailwind 4, Python 3, Claude Sonnet, Runware, YouTube Data API, Vercel, GitHub Actions
- Live: https://lawncare.center
- Custom GPT: https://chatgpt.com/g/g-69753bc916a081918cf0874f960b8f90-lawn-care-center

![Abstract editorial mark — an organic blob shape with a hand-drawn ink outline in a screen-print register and a small terracotta dot, captioned 'Figure 01 — Shape of an Org' — for the LawnCare.center case study.](https://jaketaylor.design/work/01-shape.svg)

[LawnCare.center](https://lawncare.center) is an automated content engine for a niche most people forget about until a dandelion appears. A Python pipeline turns keywords and YouTube videos into QA-scored, affiliate-linked articles. A Next.js site renders them. Research, draft, hero and section images, internal links, schema — all unattended. What stays human: whether the grass is worth the article.

Video: LawnCare.center promo — 169 lawn care guides over a grid of yard photos, the topic hub (every season, every grass, every weed), guides backed by 79 university and government sources, a checklist of cited research, scoring before publishing and 266 expert video walkthroughs, and the end card: Your yard, always ready for company. lawncare.center. (https://jaketaylor.design/work/Lawncare/lawncare-brag.mp4)

## Scope at a glance

Started November 2025 with a keyword-first generator. Five months later: two pipelines, twice-weekly ships on GitHub Actions, 74 articles across 14 categories, twelve cents per article.

**By the numbers**
- **74** Articles
- **$0.12** Per article
- **2/wk** Auto-publish
- **14** Categories
- **34** Affiliate SKUs
- **8.0+** QA floor
- **11** Python modules
- **30d** YouTube cache

**What gets read**

**Where visitors come from**

## Systems, not screens

Two pipelines feed the engine. Keyword-first: take a search term, dedupe against everything published, draft 1,500 words with Claude, score against a 30-point SEO rubric, reject anything under 8.0. Video-first: pull trending YouTube uploads, cache thirty days, have Claude evaluate the transcript, send survivors through the same draft. Both land in `drafts/` for a human look before `site/content/posts/`.

![Topic page on LawnCare.center listing 52 Lawn Health & Maintenance articles in a 3-column grid, with season filter chips for All, Spring, Summer, Fall, and Winter.](https://jaketaylor.design/work/Lawncare/article-index.webp)

*Topic page: 52 articles in one grid, filtered by season. Every card, hero, and excerpt generated by the pipeline.*

## A second surface

Every article gets a second life. I packaged everything — 74 articles, the product catalog, the QA rubric — into a [custom GPT](https://chatgpt.com/g/g-69753bc916a081918cf0874f960b8f90-lawn-care-center) that answers lawn-care questions in conversation, grounded in the same content. Personalized by location and season. Same affiliate-aware recommendations. New pieces ship; the GPT knows the next day.

Video: Screen recording of the Lawn Care Center custom GPT answering a lawn-care question in ChatGPT. (https://jaketaylor.design/lawncare/ChatGPT.mp4)

*The Lawn Care Center GPT: the same content in conversation form.*