Panel, instantly
Content
Headings and paragraphs, photographs, prices and stock, opening hours, testimonial entries, FAQ answers, the posting schedule, social links and inbox replies. All rows, all free, all live immediately.
Content engine
AI content that sounds like your site
AI-written content stops sounding like AI when every word passes through one brand voice profile before it is written.
The profile records register, sentence length, spelling and the words never to use, which is the part most tools skip, and topics come from the site’s own visibility report rather than a generic calendar. The result is that two posts written six months apart still read as the same site.
Brand voice
- Register
- Plain, warm, no jargon
- Sentences
- Short. One idea each
- Never say
- “Solutions”, “passionate”, “nestled”
- Spelling
- British
Applied to every generated word
Next up
How much does a bike service cost?
Gap found in auditDraftWinter chain care in five minutes
Keyword: chain maintenanceScheduledWhich lock actually stops a theft?
Keyword: bike lockIdea
Brand voice
Six things, and one of them does most of the work.
The banned list is the half most tools leave out, and it is the half that stops generated copy drifting into the same handful of words every generated copy uses.
- Register
- Formal, plain, warm, blunt. Whichever the site already is, described rather than guessed at.
- Sentence shape
- Long and considered, or short and one-idea-each. This is most of what makes writing recognisable.
- Vocabulary
- The words this business genuinely uses for its own work, including the unglamorous ones.
- Never say
- A hard list of prohibited marketing tropes. Phrases such as “unleash your potential”, “game-changer”, “bespoke solutions” and “empowering users” are blocked. This rule is enforced more strictly than any other, on purpose.
- Spelling and units
- British or American, metric or imperial, dates in one format. Small things, noticed when wrong.
- Claims
- What may be asserted and what may not. A site that cannot say “award-winning” says so here once.
One profile per project, applied to every module that writes. The same request produces different copy on a solicitor’s site than on a skate shop, which is the entire point.
Self-improving blog
A blog that reads its own report card.
Most AI blogging tools generate posts on schedule from a topic list, which produces volume and no direction. This one starts from evidence the site already has about itself.
The standard it is written to is the published one rather than a house theory. Google’s guidance oncreating helpful, reliable, people-first content rewards usefulness however a page was produced, and itsspam policies are aimed at content made at scale to game rankings rather than to answer anything. Starting from a real finding is what keeps a post on the right side of that line.
01
It reads the audit first
The visibility report already knows which pages are thin, which questions the site answers badly, and which topics it has no page for at all. That is a better brief than a content calendar.
02
Then the keyword list
The phrases you actually want to be found for, plus the ones the site is already close on. A post is worth writing when it is within reach, not when it is merely on topic.
03
It researches before it writes
Facts come from live sources rather than memory, and every source used is cited inline in the finished post, so a claim can be checked against where it came from.
04
It drafts in your voice
Answer-ready by default: the h1 is the question, the first paragraph answers it in two or three sentences, and the questions people ask get a visible Q&A block at the bottom.
05
A fact-check gate decides what can publish
Every claim is checked against its cited source. Anything the pass cannot support is held back as a draft for review rather than published, whether autopublish is on or off.
06
The next audit judges it
The published post is scored like any other page. A post that did not land shows up as a finding, which is what makes the loop self-improving rather than merely automatic.
Approving is not a blog-only job, so it is not a blog-only screen. Everything on the site that is waiting for a decision gathers in one review queue: posts the blog has drafted, testimonials a customer has submitted, questions and answers drawn from what the site already says. Each arrives with its words editable, so approving one after a small correction is the same two clicks as approving it as written.
Nothing in the queue is visible to the public while it sits there, and that is enforced in the data layer rather than by the panel choosing what to show. A module that collects anything from outside the admin area joins the queue by existing; when it is uninstalled, its queue goes with it.
Editing
Content changes never wait for a build.
Copy, images, prices and posts live in the project's own database and render at request time, so a saved edit is live on the next page view.
Build, minutes
Structure
A new page, a new section, a different layout, a restyle, a module installed. Placing a testimonials carousel on the home page is a build; the quotes inside it are content.
Q&A
Questions about AI content
Will the blog publish without me seeing it?
Not unless you switch that on. The default is that every post waits as a draft in the review queue, alongside anything else on the site that needs a decision, until you approve it. Even with autopublish switched on, a fact-check pass still holds back anything it cannot support against its cited sources, so an unchecked claim is never published outright.
What is the answer-ready house style?
The h1 is written as the question the page answers, the first paragraph answers it directly in two or three sentences with no preamble, and the common questions get a visible Q&A block near the bottom. It exists because answer engines quote opening paragraphs, and a page that buries its answer gets summarised badly or not cited at all.
Can I edit or rewrite a draft?
Yes, in the panel, and your edits teach the voice profile. Correcting the same thing twice is a signal the profile is wrong, not that you should keep correcting it.
Does the voice profile apply to anything besides the blog?
It applies to every module that generates text: catalogue descriptions, form auto-replies, social posts, meta descriptions. A product description that sounds nothing like the blog is the tell that a site has no voice at all.
Where does the voice profile come from?
It is drafted from the copy already on the site and then edited by you. Starting from what exists is more accurate than starting from a questionnaire, because most people describe their own tone less well than they demonstrate it.
Is AI-written content penalised by Google?
Google’s stated position is that it rewards helpful content regardless of how it was produced, and penalises content made primarily to manipulate rankings. That is the same standard a human writer is held to, which is why the useful lever here is whether the post answers a real question rather than who typed it.
Are a post’s claims checked before it publishes?
Yes. Each draft is researched against live sources rather than written from memory, the sources are cited inline, and a fact-check pass compares every claim against them afterwards. Anything it cannot support is held back as a draft for review instead of being published.
Write the voice profile once.
Everything the site generates afterwards passes through it: posts, descriptions, replies, meta text. Getting it right is half an hour that pays out for years.