The problem this solves
Producing ten blog posts is rarely the hard part anymore. The hard part is everything wrapped around the writing: choosing topics that map to real demand, keeping the drafts consistent, loading each one into the CMS with the right fields, and then telling the people who need to review them. Each of those is a separate tool, a separate login, and usually a separate person. A batch that should take a day of focused work stretches across weeks of handoffs.
So the batch stays undone. Drafts pile up in scattered docs that never reach the CMS. Reviews never start because nobody was pinged with a link and a deadline. Meanwhile the topics you researched in January are still unwritten in April, and the organic channel that was supposed to compound sits flat. The cost is quiet and cumulative: every week the posts stay unstaged is a week competitors collect the queries you already identified.
How the mission runs
- Ground the batch in real demand. The mission starts with Keyword Intelligence, mapping the query space around your topic - in the stock prompt, AI sales automation - and selecting ten targets by intent and coverage gaps. Each post gets a primary query and supporting terms before a word is written, so the batch attacks demand you can actually win.
- Draft ten answer-first posts. Each post is written against its target query with a clear structure: a direct answer up top, sections that expand it, and internal logic that keeps the batch coherent rather than repetitive. Meta titles and descriptions are drafted alongside the body so nothing arrives half-finished.
- Stage everything as Sanity drafts. The agent creates each post as a draft in Sanity, mapping title, slug, body, and meta fields to your schema. Drafts stay drafts - nothing goes live until a human publishes it. If a field mapping looks off, you see it in the Sanity draft before any reader ever could.
- Post the review links to Slack. When the ten drafts exist, the mission posts their links to your #content channel in one message, with a line on what each post targets. The review conversation starts where your team already lives, with the actual drafts one click away.
- Open a review task per post in Asana. Each draft also becomes an Asana task carrying the Sanity link and the target query, so review has an owner and a due date instead of relying on someone remembering the Slack thread. The batch is done when all ten tasks are closed by humans who read the work.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
Ten complete blog post drafts sitting in Sanity with meta fields filled, a single Slack message in your content channel linking all ten, and ten Asana review tasks - one per post - each pointing at its draft. Your team wakes up to a reviewable batch with a workflow already attached, and publishing becomes a sequence of approvals rather than a production project.
Make it yours
- Change the topic and count: five deep pillar posts on your core category instead of ten, each targeting a head term with supporting sections.
- Route the links to a client-facing Slack channel and assign the Asana tasks to their marketing lead for agency-style review.
- Add a cadence: have approved posts publish one per weekday so the batch drips out instead of landing at once.
Frequently asked questions
Will anything publish without my review?
No. Every post lands as a Sanity draft, and the Slack links plus Asana tasks exist precisely so humans review before anything ships. Publishing is a deliberate action your team takes in Sanity, or a follow-up mission you explicitly authorize after approvals.
How does the agent match our brand voice?
Point the mission at your style guide, a few representative posts, or both, and the drafts follow that register. The first batch is also a calibration pass: notes you leave in the Asana review tasks feed the next run, so voice tightens with each cycle.
What if our Sanity schema is heavily customized?
The mission maps to the document type and fields you name, and anything it cannot map cleanly stays visible in the draft rather than silently dropped. Most teams run one test post first, confirm the mapping, then let the full batch flow through the same path.