This is the first post on this blog, and it exists because of a failure. So let me start there instead of with a mission statement.
For sixteen months I ran a content machine on a side project called Startups Españolas. It published on its own, twice a day, no human in the loop. By the time I looked hard at the numbers it had put out around 2,300 articles. I was quietly proud of the pipeline. It was a real piece of software: a Laravel backend, prompts stored in the database, several LLM providers wired in with fallback, automatic publishing straight to the live site. It felt like leverage. Then I pulled the Search Console data for the full run and sat with it for a while.
Here is the monthly click trend, straight from the GSC API, nothing rounded in my favor:
| Month | Clicks | Impressions | Avg. position |
|---|---|---|---|
| 2025-07 | 638 | 76,194 | 33.2 |
| 2025-09 | 695 | 56,458 | 19.8 |
| 2025-10 | 634 | 33,378 | 13.5 |
| 2025-12 | 428 | 31,475 | 16.8 |
| 2026-02 | 166 | 15,838 | 23.2 |
| 2026-04 | 115 | 14,969 | 22.1 |
| 2026-06 | 69 | 9,236 | 32.1 |
Peak was September 2025 at 695 clicks in a month. The last full month, June 2026, did 69. That is a 90% drop. Average position over the same stretch went from roughly 13 back down to around 32, which in plain terms means Google moved my pages from the bottom of page one to the bottom of page three.
The part that actually stung was the long tail. Over a 90 day window, 1,567 of the pages picked up impressions, so Google had indexed most of the machine's output. But only 147 pages, about 6%, got a single click. I had built two thousand articles to feed a hundred and fifty that did anything, and even those were fading.
The story I had been telling myself was "more articles, more surface area, more chances to rank." That is intuitive and it is wrong. More thin pages did not add up to more traffic. They competed with each other for the same weak queries, diluted whatever topical signal the domain had, and gave Google a large pile of near-duplicate, machine-written content to judge the whole site by. That is exactly the kind of thing the helpful-content updates are built to push down.
The traffic that did exist was not coming from the volume at all. It came from a small handful of genuinely useful startup profiles that happened to rank. The machine was not the asset. Those few real pages were. I had spent the effort on the wrong half.
If the slow decline was the lesson, the fast one was the wake-up call. The brand was Spanish, es-ES only, on purpose. One day a channel without a language guard generated a batch of posts in English and pushed them live, automatically, onto a Spanish-only site. No human saw them before they were public. Nothing catastrophic happened. But that is the whole point of automation without a human gate: the day something does go wrong, it is already live before you know it exists. A pipeline that can publish a good post on its own can publish a bad one on its own, and it will, eventually, at the worst possible moment.
I am not anti-automation. I let AI write most of my code and I am not going to pretend otherwise on a blog about building with it. The lesson is narrower and more useful than "AI content bad":
This is the opposite bet. It is static, hand-written, and English. There is a small script that gathers what I actually did across my projects each week, git history and working notes, and hands me the raw material. Then I choose one thing worth writing about and write it myself. I read every post before it goes live. There is no CMS, no queue, no provider zoo. When you are reading this, a human decided it was worth your time first.
I build around a dozen real projects mostly solo, with AI doing a lot of the heavy lifting, and I am going to write honestly about how that actually goes. Some of it works. Some of it is this post. Both are worth showing.
If that sounds useful, the RSS feed is in the footer. No newsletter popup, no cookie wall. Just the work.