✍️ A content engine that ships weekly

The outcome: a standing pipeline that turns “we should blog more” into a weekly rhythm: the research arrives Monday, the outline waits for your 30-second nod, the draft is ready Wednesday, and your total involvement is one approval and one edit pass. Consistency - the thing that actually compounds in content - stops depending on your motivation.

The division of labor that works: agents are excellent at research, structure, first drafts, and consistency. You are irreplaceable for taste, stories, and the opinions only you can have. This playbook splits exactly along that line - and the ticket flow is the seam.

1. Write the voice doc once. Ask your content agent: “Interview me for 10 minutes about our content voice, audience, and the opinions we hold that competitors don’t. Write it as a doc called Content Voice.” Every future piece cites it.
2. Set the weekly schedule (content agent → Configure → Scheduled Tasks): Monday 7 AM, instructions below. This is the heartbeat of the whole engine.
3. Approve outlines from tickets. Monday’s run ends with a ticket: three angles, one recommended. Picking one takes 30 seconds and is genuinely your call - that’s why it’s a ticket and not a guess.
4. Edit the draft, don’t rewrite it. Wednesday’s draft lands in Docs. Do one pass for taste and stories. If you’re rewriting structure, fix the outline step in the schedule instead - that’s a briefing bug, not a draft bug.
5. Add distribution to the same schedule once publishing is steady: repurpose into a thread + newsletter section, schedule the sends. One piece becomes four artifacts without a second brief.

The Monday schedule instructions

Every Monday: research what our audience [describe them] argued about last week — forums, communities, competitor blogs, anything real. Cross-reference our Content Voice doc and past posts in Docs so we never repeat ourselves.

Produce: three post angles, each with a working title, the argument in two sentences, and why now. Recommend ONE. Open a ticket with all three for my pick — do not start drafting until I choose.

After I pick: write a full outline, then a complete draft in our voice, saved to Docs as “Draft — [title]”. Cite sources inline. Open a second ticket when the draft is ready for my edit.

💳 Cost note: The research step uses credit-metered tools (web search, page scrapes - a few credits each; the exact menu is in Billing → scrape pricing reference). A weekly research run typically lands in the tens of credits. Watch one week in Runs → the run detail shows credits per research call.

💡 Tip: The quality lever is the voice doc, not the prompts. When a draft misses, don’t fix the draft - add the missing rule to the voice doc (“never open with a rhetorical question”). Every future draft inherits the fix. This is what “the squad remembers” means in practice.

What good looks like, week four: you’ve published four weeks straight. Your Monday ticket takes 30 seconds, your Wednesday edit takes 10 minutes, and Docs holds a growing library - voice doc, published posts, and research files your agent now cites back to you. The engine survived two weeks where you were too busy to care. That’s the point.