Byline finds the keywords, writes every piece in your brand voice, and publishes it straight to WordPress, Webflow, Shopify or Sanity. You never open a keyword tool, and you never write a brief. If you’d rather check each piece before it goes live, there’s a switch for that.
You do not have a traffic problem. You have a nobody wrote the page problem. Search is how people look for what you sell, and they are not searching for your name.
Nobody types your company into Google unless they already know it. They type the thing they are trying to fix. Whoever wrote the page that answers it gets the visit, and right now that is somebody else.
Google and ChatGPT can only recommend pages that exist. A homepage, a pricing page and an about page give them almost nothing to work with, so they hand the answer to a site that has written more.
The gap does not hold still. Every article they publish is a question they now own and you do not, and they are still publishing while you read this.
A page written this month is still bringing people in two years. The one you did not write is still bringing nobody, and in two years it will still be bringing nobody.
Every visitor arrives because you sent them. Search has almost nothing of yours to offer anybody who has not already heard of you.
30 a month on Starter is 90 questions answered by the end of a quarter, each one a page somebody can land on without knowing your name first.
The question was never how to publish more cheaply than a freelancer. It was whether anything gets published at all.
Most “AI writers” are a text box. You paste a prompt, you get 900 bland words, and then you still do the actual work: checking what already ranks, fixing the structure, adding the links, making the image, writing the schema, pasting it into your CMS, remembering to do it again next Tuesday.
Byline is the opposite arrangement. It is a full production line, wired end to end, and it starts before you've had an idea. Point it at your website and it reads what you sell, what you've already covered and what you're missing, then pitches the stories itself. A published, structured, internally-linked, illustrated article comes out the other end, with a scorecard proving it cleared the bar before it printed.
And because it's a newsroom, you don't manage prompts. You manage a publication: a schedule, a story desk, a quality standard, an archive, and a circulation report. That's the whole idea.
Link your site once. Byline reads your pages, infers what you actually do, and discovers the keywords worth chasing, grouped into clusters, checked against what you've already published, and pitched to you as stories. Then it reads the live search results before writing a word.
Each section is drafted against a brand voice you define once (vocabulary, sentence rhythm, things never to say), then woven with internal links scored across every page you have already published. Figures are taken from checked sources, never invented to fill a sentence.
Nothing prints below your quality bar. Cleared articles get JSON-LD schema, a generated hero image with alt text, and a push straight to WordPress, Webflow, Sanity or your own endpoint, on the cadence you set and forget.
Four movements, one continuous run, and the first one happens without you. Byline reads your linked site to decide what to write about, researches before writing, and judges the result before anything ships.
Every draft is scored before it can publish, and a figure that cannot be traced to its source is sent back with it. Fall short and Byline revises itself. Still short, and it's spiked and held for a human. Your name only goes on work that earns it: that's the entire point of a byline.
Every tool in this category promises human-sounding output. Most of them are a text box and a prompt that begs the model to try harder. Here is what Byline does instead, and where it still falls short.
Vocabulary you use, words you would never say, sentence length, how formal, whether the company says “we” or “I”. Every section is drafted against that profile, then graded on how close it landed. Voice drift is part of the gate, not a footnote.
Em dashes, “delve”, “in today’s fast-paced world”, the hollow closing paragraph that asks what you think. A sanitiser removes them after the model is finished, because asking a model not to write something and it not writing it are different events.
Every number is checked against the page it was pulled from. If the source does not contain the claim, the draft goes back for revision. One invented statistic is enough to lose a reader for good, and it is the failure mode that gave AI content its reputation.
The copydesk scores structure, depth, voice match and sourcing. Short of the floor, Byline rewrites it. Short again, it is held for you instead of published. You can raise the floor, or hold everything for approval.
Byline cannot interview your customer. It does not know the thing that happened in 2019 that made you start the company, and it will not invent one. Original experience is yours to add, and on your five or six most important pages you should. What Byline handles is the other ninety questions, the ones you were realistically never going to write about, at a standard that holds up when somebody lands on them.
Google’s stated position is that it ranks useful content and does not care what produced it. Sites get hit for publishing thin pages at volume, not for using a machine. That is the entire reason the floor exists. Thirty articles a month that each clear 80 is a different thing from three hundred that clear nothing, and Byline is built to do the first one.
Read one it wrote →Byline isn't one screen with a Generate button. It's a publication you run: six desks that each answer a different question about your content operation.
The centrepiece. Watch a story typeset itself in real time: copy appearing paragraph by paragraph, the copydesk scoring it live, then a To Press or Spiked stamp landing on the galley. You see exactly what was written and why it cleared.
Your morning edition. A masthead of the numbers that matter (published this week, currently in production, held for review, keywords ranking) over a story index of everything moving through the building right now. One screen tells you whether the paper is on schedule.
Every article ever set, filed as a dense archive ledger: slug, keyword, word count, quality score, status. Filter by status, sort by score, find the thin piece from March and send it back through the press for a rewrite.
Where the paper decides what to cover, and it fills itself in. Byline reads your linked site and returns keywords already grouped into topical clusters with volume and difficulty, framed as editorial pitches rather than a spreadsheet. Green-light a cluster and every keyword in it becomes a queued story.
Set the cadence once (three a week, one a day, Tuesdays only) and the press runs itself against the calendar. Every planned edition sits on its date, and you can add one to any day or pull it off again.
Did it work? A plain editorial report on readership: organic sessions, ranking keywords, impressions per story, position movement. No vanity dials, just the numbers that tell you whether to raise the cadence. This desk is still being built. Rank tracking is live today; the readership figures arrive with the Search Console wire.
Byline reads from the search wires and prints to your stack. Connect once in Settings; the presses handle the rest.
One flat monthly rate: no per-article billing, no usage meter, no surprise invoice at the end of the month.
No. Link your website and Byline reads it: what you sell, who you serve, what you've already published, then discovers the keywords worth chasing and groups them into clusters on the Story Desk. You approve what you like, or let it run on autopilot and pick for itself. You can always add your own keywords on top.
No, but you can. Byline only auto-publishes work that clears your quality floor; anything below it is spiked and held. Many teams run fully hands-off on cleared work and only open the pieces that got held. You can also set the floor higher than 80, or require manual approval on every article.
Only where you point it. You connect a CMS in Settings and choose whether Byline publishes live, publishes as draft, or just hands you the finished package with structured data and image attached. You can disconnect at any time and the archive stays in the Library.
One flat monthly rate for your article allowance. There is no per-article charge, no usage meter and no overage bill: what you see on the plan is what you pay. Change plan or cancel any time.
Yes. 80 is the default floor; you can set it higher, require manual approval on every article, or hold specific clusters for review while letting the rest run hands-off.
Paste a domain. Byline reads the site, works out which searches it has the authority to win, and scores what is missing. About a minute, nothing to install.