Nobody using GetCited has spent an hour on SEO since we launched

The same three days, every month

Crawl the site. Read the findings. Twelve titles are too short, nine pages have no canonical, nothing carries a modified date, and there is a 404 that nobody linked on purpose. Fix them. Then the other half: ask the engines the questions your buyers ask, write down who they cited, notice that Perplexity picked you up and ChatGPT did not, and work out which page to make retrievable.

None of that is hard. All of it is a Tuesday. And next month it is Tuesday again, because the site changed, the engines changed their minds, and the competitors published.

The thinking is the small part

There is real judgment in SEO: what your product is for, which page deserves to exist, what you are willing to claim. That part is yours and it takes an afternoon a quarter. The rest is bookkeeping with a search engine attached, and it is the rest that never fits in the week. It gets postponed, then batched, then done badly in a hurry, which is how a site ends up with a title tag that says nothing and a 404 that has been live since March.

There is now a worker for exactly this

Your coding agent already has the context that used to make this work expensive. It knows the codebase, it knows your components, it ships through the review you already run, and it is available at 3am on a Sunday. What it could not do was see the site the way an engine sees it: it cannot crawl your production HTML, ask five answer engines your questions, or track where you rank tomorrow.

That is the half we built. GetCited measures and reports; the agent decides and fixes. We never write your copy or your code, and that is deliberate, not modesty: we have no idea what your product is, and the agent with your repo open does.

The handoff is one sentence

You connect the client you already use, then type the sentence. What happens next is the loop above, without you in it:

• We crawl the site the way an engine does, JavaScript not executed, robots.txt respected.
• Every finding comes back with the measurement that fired it: the character count, the missing tag, the page that answered 404.
• Your agent fixes what it agrees with, in your repo, and says why when it refuses.
• The next crawl closes the ones that are actually fixed. There is no 'mark as done' button, because that records an opinion where a crawl records a fact.

Which brings us to the number

Since the platform went live, the people on it have not opened a keyword tool, read a crawl report, or spent an evening on title tags.

What zero hours does not mean

It does not mean no work happened. It means the work moved: the agents did it, and we measured what changed. It does not mean instant rankings either. Position moves on the engines’ schedule and answers are stochastic, so we report frequency over a window and only call a gap a gap after three runs agree.

And the autopilot still waits to be asked. Scheduled runs and a digest that lands in your inbox when something moves are what we are building next; today the agent goes when you say go. We will publish what that changes, including the months it changes nothing.

The three days did not disappear

They stopped being yours. That is the whole product, and the rest of the site is just the evidence for it: how each number is measured, and what our own audit found when we pointed it at ourselves before anyone else.