New feature launch: Scribe our onsite writing agent
If you caught the Conversation Radar post, you know the premise: AI engines cite people talking about brands, so we built a radar for those conversations.
But there's a second half to AI visibility, and it's the one brands keep asking us about: the content you publish yourself.
Here's the honest version. Owned content works harder to earn a citation than third-party content does. We've tracked 13,000+ citations across ChatGPT, Perplexity, Gemini, Copilot, and Claude, and brand articles DO show up in that set. The ones that do share a shape: they answer one specific query head-on, they carry real facts, and they cover the query from multiple angles, not one post.
Since the beginning, our content team has been building exactly that for client brands by hand. One primary article, three supporting ones per query. It worked, but it doesn't scale past the clients we can personally serve.
So we turned the system into an agent. Meet Scribe.
You pick a query, ideally one where AI engines never mention you (the dashboard sorts your queries by least visible, which is a mildly painful but useful view). You paste a few URLs about your brand so Scribe writes with your facts instead of guessing. Then it writes the full set: 1 primary + 3 supporting articles, each with metadata, ready to publish on your blog.
It doesn't publish for you, and it doesn't pretend a first draft is final. You review, edit, ship. Next versions add in-dashboard editing and Scribe asking YOU questions before writing, because a content agent that never asks anything is just a template with confidence.
Question for this community: if an agent drafted your company blog, where's your line? Full autopilot, human-reviews-everything, or "never, blogs should be handwritten"? The answers genuinely shape what we build next.
Check out how this works here - https://scribble.network/feature/scribe
If you're curious about how to use scribe you can check out the tutorial here -
Replies
someone on my team has been looking for a better way to plan supporting articles around important search topics. I think this could be worth showing them, especially because the content is organized around a specific query rather than random blog ideas.
The part about Scribe asking questions before writing caught my attention. I think my biggest concern with AI-written content is when it sounds confident without knowing enough about the business. Giving the agent a chance to ask for context could make a big diffrence.
my prefrence would definitely be human review before anything goes live. AI can handle the first draft well, but I still want my own voice, examples, and opinions in the final version. The review step here feels important.
The fact that you're using actual brand URLs as context is probably one of the most important parts. Otherwise these agents can convincing while quietly inventing details.
The interesting part here isn't generating content. it's generating content around queries where you actually have a visibility gap. That makes the workflow feel much more stretegic than a normal AI writer.
For me it's definitely human reviews everything. The agent can do the heavy lifiting, but the final voice , claim, and example should still come from the people behind the brand.
I’m somewhere between human review and autopilot for low risk educational content, but anything making claims about the company would need a proper review.
I like that you're not pretending the first draft is publication ready. That expectation alone makes me much more interested in trying something like this.
The human review step is important to me. I'd comfortable letting an agent handle the research and first draft, but I'd still want a person to verify facts and add the company’s actual perspective.