Saving an article does not publish it. If your site content lives in a
connected GitHub repository, you can separately ask daoco to propose the
article as a pull request. You still review and merge that pull request.
Ask for an article
1
Ask for an article in brand chat
“Write an article about X”, “write a blog post that beats this page”, or
paste a URL you want to write against.
The Articles tab
Open Home, choose Website, and open the Articles tab. Every article the brand has is listed there with its title, when it was last updated, its word count, its version, and its publication state: Draft, In review, Unverified, Merged, Published, Published · unpublished changes, or Archived. See Publication for what each one means. The first time you open it, the tab offers to bring in what you already have. Enter the page where your articles live, for exampleyoursite.com/blog, and choose Bring in my existing articles. daoco reads them and saves each one as an article here. Draft a blog post starts a new one instead.
Once the tab has articles, Draft another starts the next one. If the last run finished without producing anything, the tab says so and links you to that thread.
Import existing articles
Ask “Import the existing articles from example.com/blog into my article library.” Daoco discovers article candidates through the site, sitemap, and linked feeds, then saves each readable article directly from its HTML. It converts the body to Markdown without asking a model to rewrite it. Source bylines and dates are recorded when present; missing dates are not guessed. The saved article holds the full extracted body. Chat and panel previews may be shortened; Show full article loads the saved body. Pages without an identifiable article body, unreadable pages, and articles exceeding the size limit are skipped with a reason rather than saved as partial copies. Discovery is bounded and reports when it cannot establish a complete inventory. Retrying an import reuses the article for that brand and canonical URL and preserves any edits you made. Older imports without extraction metadata are reported as unverified, not silently replaced. The final report links the saved articles and lists existing, skipped, or incomplete results. An imported source URL is a link, not a publication receipt. Imported articles show Unverified until a publication check matches a saved version to the destination.Writing against a real page
Give daoco a URL and it reads the page before it outlines. It crawls the page and uses what is actually there (the real headings, the real length, whether the page has a comparison table or an FAQ, what schema it carries) to decide what your article has to match, where it can go deeper, and what page one is missing. That is the difference between “write me a long article about email deliverability” and an article built to beat a specific page. Without a URL, daoco writes from your brand context and says so rather than implying research it did not do. The panel lists the pages daoco reports auditing for the brief, under a heading that says exactly that. It is the writer’s own account of what it read, not an independently verified record, and when nothing was reported the panel says so rather than leaving the space blank. Those audited pages are research provenance, not automatic article citations. A competing guide that helped shape the outline stays in the panel and does not appear in the article’s Sources section unless the article quotes it or uses it to support a factual claim. Sources that do support a claim remain in the article so readers can verify it.What gets saved
Title and slug
The headline plus a URL-safe slug, derived from the title unless you name
one. Collisions inside a brand are resolved automatically.
Body
The full article rendered with its headings, lists, links, quotes, and code.
Copy the original markdown out of the panel with one click.
Search metadata
Meta title, meta description, the single target query the article is written
to win, and any secondary queries the outline genuinely covers.
Version history
Every revision appends a version with a one-line summary of what changed.
Nothing is overwritten.
Revising
Ask for changes in the same thread (“tighten the opening”, “add a pricing section”, “retarget this at b2b email deliverability”) and daoco revises the article in place. Each content change appends a version; the panel shows the history. You can continue in the article’s chat thread. Ask “open a pull request for this article” to send the saved article to your connected GitHub repository. You do not need to ask for another writing pass. See Connect GitHub for the approval flow. Every article appears under Website > Articles, drafts included. A published article opens its live page in a new tab; an article in review opens its pull request; a draft opens the chat thread where it was written, which is where editing happens. Metadata-only edits are recorded as such, so “we rewrote it” and “we retitled it” are different entries.Publication
Writing and publishing are tracked separately. An article carries a publication record that says where it lives and what has been confirmed there, and the label you see is derived from that record, never from anyone saying so.1
Link the article to its destination
When daoco opens a pull request for an article, it records the repository, the branch the site deploys from, the article’s file, and the pull request. When an article you already have is live somewhere, tell daoco the page (“this article is published at yoursite.com/blog/x”) and it links that page to the saved article instead of importing it again, which would create a duplicate. Articles you brought in with Bring in my existing articles are linked to their live pages already.
2
Check it
Open the article panel and choose Check now, or ask “is the article live yet?” in chat. daoco reads the destination and the live page and matches them against the article’s saved versions. For a repository, it reads the file on the deploy branch: a match there means Merged. Only a matching live page makes the article Published. If the page is missing right after a merge, the site has not deployed yet; check again later.
3
Watch for unpublished changes
A successful check pins the version that is live. If the article is revised afterwards, the label becomes Published · unpublished changes and the panel names both versions. Publishing the change is a new pull request or CMS update, followed by another check.
- Draft: no destination recorded.
- In review: linked to an open pull request; nothing confirmed yet.
- Unverified: linked to a page or CMS record that has not been checked.
- Merged: the file is on the deploy branch, but the live page has not confirmed it.
- Published: the live page carries the article. CMS records are not read directly; the page is the evidence.
- Published · unpublished changes: the live copy is an older version than the saved article.
Indexing
Whether Google has the page is a third fact. With Google Search Console connected, Check indexing in the panel (or “is it indexed?” in chat) asks Google directly and records the verdict, coverage state, and last crawl on the article. A live page that is not indexed yet is normal for the first days after publishing.Archiving and restoring
Articles you no longer want are archived, not deleted. Use Archive in the article panel, or ask in chat. Archived articles stay queryable and can be restored from the same button.Current limits
- No direct publishing. Articles are saved in daoco. daoco can propose a reviewed article as a pull request to a connected GitHub repository, but it does not merge the pull request or publish directly to WordPress or Webflow. It verifies publication by reading the repository and the live page; CMS records themselves are not read yet, so a CMS-published article is confirmed from its live page.
- No keyword data yet. daoco grounds a brief in the page it crawls, not in search volume or difficulty. This is true even when your workspace has keyword data and Search Console available to the SEO agent; the writing lane does not read them yet. It will tell you the target query is inferred rather than measured, and it will not invent numbers.
- One article per piece. Revise the article you have rather than asking for a second one on the same topic; the history is more useful than two drafts.