Workflow · Blog & Article Drafting

Blog drafts that don't read like AI slop.

Because you write the brief. You set the voice. The workflow drafts. It never publishes. Real sources cited inline. Drafts land in your editor's queue. Exports to Sanity, Webflow, Ghost, or WordPress.

Take the AI Assessment Request a Sample Draft

3–5x

Output. Same editor bar. Same voice. More drafts to choose from.

Auto-publishes. Every draft goes through you.

~30 min

From brief to first draft. Sources attached inline.

Why most AI content fails

Most AI Content Workflows Make Slop

Here is what goes wrong on almost every "AI blog" build we see.

No real sources

A prompt-only generator writes from what the model already knows. That sounds fine. It is not. The model fills in gaps with plausible-sounding sentences. Some of those sentences are wrong. When a reader checks the citation, the source does not exist. Trust gone.

Voice drift

The first draft sounds like you. The fifth one sounds like every other AI blog on the internet. Without a tuned voice guide, AI defaults to the middle of the bell curve. Generic. Forgettable. A regular reader will notice in 30 seconds.

No human review step

Tools that promise "AI publishes for you" sound like a win. They are not. You lose the editor's eye. You lose the moment where someone notices a wrong fact, a weak hook, a flat opening. Cheap content is not the same as good content.

It goes live looking like everything else

Same hooks. Same structures. Same conclusions. Google notices. Readers notice. You end up with a content library that exists but does not work. You started this to grow the business. Now you are blending in.

How It Works

Five Steps. You Drive Two of Them. AI Drives the Boring Three.

You stay in the loop the whole time. Nothing publishes without your sign-off.

You write the brief

You tell us the topic, the reader, the point of view, and what you want them to do after reading. Short brief. 10 minutes. We give you a template.

Research agent pulls real sources

A research agent pulls 10 to 30 real sources. It picks the strongest ones. It cites them by URL. You can check every claim. No made-up citations.

Draft built to your voice guide

The draft is written to match a voice guide we tune from your real writing. Same words you use. Same rhythm. Same way you open and close. Not the model's default voice. Yours.

Sources cited inline

Every claim that needs a source has one, inline. Your editor can check the work in one pass. No hunting. No guessing.

Lands in your editor's queue

The draft shows up in your CMS as a draft post. Or in a doc your editor reviews. You publish when you are ready. The workflow never publishes for you.

Honest about AI

Where This Helps. Where It Is Just Okay. Where You Still Drive.

AI is good at some parts of writing. It is bad at others. We will not pretend otherwise.

Where it helps

Hunting sources and citing them. Building a first-draft scaffold. Staying on-voice after we tune it. Generating A/B test variants. Transcribing meetings and pulling out highlights.

Where it is just okay

Strong opinions and hot takes. Real stories from your life. Humor and timing. The inside-industry stuff the model does not know. We let AI try. Then you fix it.

Where you still drive

The point of view. Real interviews and original reporting. The call on what to publish. The voice itself. These do not belong to AI. They never will.

The Stack

Built on Tools You Already Use (or Will)

Nothing here is custom magic. We wire best-in-class tools together. You own the prompts and the integrations.

Claude GPT-4 Sanity CMS Webflow Ghost WordPress Notion Google Docs Your voice guide Your editor's queue

Common Questions

What Marketing Teams Ask Before They Start

If it reads like AI, yes. If it reads like you (because you wrote the brief and reviewed the draft), no. Google's guidance is about helpful content for readers, not about whether AI was in the loop. Sources are cited inline. We do not make up fake citations. Your editor is the last set of eyes before publish.

Then we tune the voice guide. Usually 2 to 3 rounds before drafts feel like you. We start by reading 10 to 20 of your existing posts. Then we build a voice guide that captures how you actually write. If something is off, we fix the guide. The guide gets better over time. Drafts get closer to yours over time.

Those are tools you type a prompt into. This is a pipeline. Research agent. Voice tuning. Editor review. CMS export. All wired together for your team. You can still write a one-off draft in Jasper or Copy. This is for teams that want to produce consistently, in their voice, over time.

Sources are cited inline. The draft is written from those sources, in your voice, not lifted from them. If you want, we can run a plagiarism check as the last step. We have not had a case where a draft tripped one. But we are happy to add the check if your legal team wants belt-and-suspenders.

Yes, but with a caveat. The research agent pulls real sources. The draft is written from them. For deeply technical or niche topics, you (or a subject expert on your team) will want to edit the draft. We do not pretend AI knows your industry better than your industry experts do.

Two to three weeks. Week one is voice tuning. We read your existing content. We build a voice guide. Week two is wiring up the research agent and the CMS export. Week three is iteration. First draft usually ships during week two. We tune until it sounds like you.

No. The writers we work with end up doing more interesting work. The boring 60 percent of the draft is the AI's job. The hook, the angle, the original reporting, the voice, the edits — those are still your writers. Most teams ship more content and have better content because their writers are no longer buried in source-hunting.

That is the whole point. The first thing we do is tune the workflow to your voice. Your hooks. Your sentence rhythm. Your endings. Your words. Drafts that do not sound like you are drafts you can't ship. We do not pretend otherwise.

No. Your content is yours. Your prompts are yours. Your voice guide is yours. You can take the whole workflow with you if we ever part ways. No vendor lock-in.

Build cost depends on your CMS, your voice complexity, and how many sources we wire in. Plus a light monthly cost to keep it running. Send us your monthly content target and your current CMS — we'll come back with a real number in 3 business days.

Two Ways to Start

Take the AI assessment for a structured read. Or send us a topic and a few links to your current writing. We will come back with a sample draft in your voice.

Self-serve · 8 minutes

Take the AI Assessment

A short structured read on where AI fits your content operation. Tailored plan in your inbox.

Talk to us

Request a Sample Draft

Send us a topic and 3 to 5 links to your current writing. We will come back with a sample first-draft article in your voice. Plus a build plan and timeline.