Content Machine Playbook

How to produce a full week of content — blog posts, Twitter threads, LinkedIn posts, email — in a single overnight session using ZELDA and the SSBAA content pipeline. Templates included.

What this guide covers

ZELDA produced four complete content files in a single overnight session on Day 1 — blog posts, a LinkedIn article, Twitter threads, and a content calendar. This guide documents exactly how we structured that workflow so you can replicate it.

The core idea is simple: content creation at volume requires a clear brief, a structured pipeline, and a review loop. Without the brief, the agent produces generic content. Without the pipeline, outputs pile up without a publishing schedule. Without the review loop, you're publishing unreviewed AI output, which is how you end up with a hallucinated agent as a blog author.

The content pipeline

The pipeline has five stages. Each stage has a defined input, a defined output, and a defined owner. Nothing moves to the next stage without the previous one being complete.

1. Brief (RAY, Sunday evening) — a structured document covering the week's themes, key messages, target audience, any specific angles or stories to use, and what not to say.
2. Research (ZELDA, 22:30) — ZELDA reads the brief, scans competitors, identifies current conversation in the space, pulls relevant data points, and writes a research summary before touching any content.
3. Draft (ZELDA, 23:00 onwards) — all content drafted from the research summary. Blog post first, then adapt down to LinkedIn, then Twitter threads, then email. Same core ideas, different formats.
4. Review (YOSHI, 06:00) — YOSHI checks all outputs against the brief: on-message, correct author attribution, no fabricated claims, no hallucinated agents. Flags anything that needs RAY's attention.
5. Publish (RAY, 08:00) — RAY reviews YOSHI's flagged items, approves the rest, and publishes to the relevant platforms.

Writing the brief

The quality of ZELDA's output is directly proportional to the quality of your brief. A vague brief produces generic content. A specific brief produces content that sounds like you wrote it.

Content calendar automation

ZELDA produces a weekly content calendar every Monday at 09:00 as a cron job. The calendar maps each piece of content to a day and platform, ensures variety across the week, and prevents the same topic appearing on the same platform twice in a row.

Platform formats

The same core idea adapts to each platform. ZELDA knows the format rules for each — you just need to tell her which platforms you're targeting in the brief.

Brand voice consistency

The biggest risk with AI-generated content is generic output that sounds like every other AI-generated content. ZELDA avoids this when the brief is specific, but the brand voice rules also need to be written into her system prompt.

Here are the exact rules we use in ZELDA's SOUL.md file:

The review loop

YOSHI reviews all ZELDA outputs before anything is marked ready to publish. The review checklist:

• Does it match the brief's core message?
• Is the author attribution correct? (RAY on all public content)
• Are any statistics or claims verifiable?
• Does any agent name appear that shouldn't? (Hallucination check)
• Is the CTA present and correct?
• Is the word/character count within the platform spec?

YOSHI flags anything that fails a check and includes it in the morning brief with a specific note. RAY reviews flagged items before publishing. Everything that passes goes into the approved queue.

Scheduling and publishing

We use Typefully for Twitter scheduling — ZELDA writes the threads, RAY pastes them into Typefully and schedules. LinkedIn posts go directly. Blog posts deploy via the Next.js content pipeline when FOX merges the MDX files.

What to measure

Content without measurement is guessing. The three metrics we track weekly:

• Output volume — did ZELDA produce everything on the calendar? If not, why not?
• Engagement rate per platform — which platform is performing best this week vs last?
• CTA clicks to ssb-aa.com — which piece of content drove the most traffic?

ZELDA produces a one-page performance summary every Monday alongside the new content calendar. It covers last week's top performer, lowest performer, and one recommendation for this week.