- Home
- How it works
How we run your website without touching your app.
Your site stays in your own GitHub repo. We change only the folders you've agreed to, and you get an email every time we do.
Getting started
From your first message to the first article.
Six steps, in the order they happen.
You tell us your site
Fill in the form with your website and where the code lives. We check that the site is in a GitHub repo and that we can work with how it's built. We reply by email. No payment is taken at this stage.
You join and give us access to one repo
Access covers the repo your website lives in and nothing else. You can remove it whenever you like, from your own GitHub settings.
You send us the facts
What the product does, who it's for, what it costs, who you compete with, and which claims we may make. We write only from what you've given us. If a fact isn't in your brief, we ask.
We read the repo and propose a list
We find where your marketing pages and content live, and send you a list of the folders we'd work in. You confirm it once.
We take a baseline
A technical check of the site, a record of what AI assistants say about your product today, and a map of the questions your buyers ask. Later reports are measured against it.
The work starts
Four articles a month, fixes as we find them, a check every day, and a report every week.
The fence
What we can change, and what we can't.
Most new products keep the website and the app in the same repo. We don't ask you to split them. We agree three groups of files, and a check runs before anything is sent to your repo.
Articles are content files, never code, which is why we publish them without waiting on you.
If your landing page is tangled into your app code, our first job is one pull request that moves the marketing pages into their own folder. That one needs your approval, because it changes structure.
- your-product/
- content/blog/We publish
- public/blog-images/We publish
- app/(marketing)/You approve
- app/sitemap.tsYou approve
- app/robots.tsYou approve
- app/dashboard/Blocked
- app/api/Blocked
- lib/auth/Blocked
- prisma/Blocked
- .envBlocked
- package.jsonBlocked
| Group | What's in it | What happens |
|---|---|---|
| Content | Blog posts, article images, content folders | We publish once the preview build passes |
| Pages | Marketing pages, sitemap, robots file, page titles and descriptions, structured data | We open a pull request and wait for your approval |
| Everything else | App screens, API routes, login, database, billing, environment files, dependencies | Blocked. If a change touches any of it, nothing is sent |
Week to week
What happens once you're up and running.
Every day
We check that your pages load, that crawlers can read them, and that nothing has been blocked or broken. Anything we find goes on your task list.
Every week
You get a report by email: what changed on your site, what moved in search and in AI answers, and what's next.
Every month
Four articles, chosen from the questions your buyers ask and the gaps we find. Comparison and alternatives pages count among them.
Whenever you ship
Tell us when you launch a feature or change a price. We update the pages that mention it, with your approval.
Straight answers
What we won't promise.
Nobody controls Google or ChatGPT, so we don't guarantee a ranking or a mention. We commit to the work, measured against your baseline and reported every week.
We also won't add things that have no measured effect and call them progress. Google stopped showing FAQ rich results on 7 May 2026. Ahrefs tracked 1,885 pages that added structured data and found no major uplift in AI citations. Google says an llms.txt file has no effect on its search.
The sources for all three are in How AI assistants choose which products to recommend.