Content Production Workflow — Advisory OS

This is the repeatable weekly process for producing a campaign and assembling it into an evergreen conversion asset. All production happens in Claude Code.

The Pipeline

Transcripts + JSON
        ↓
Discovery SOP → Concept Brief (the spark)
        ↓
Interactive Article (the narrative)
        ↓
Micro Tool 1 + Micro Tool 2 (the self-diagnosis)
        ↓
Briefing Script + Walkthrough HTML (the behind-the-scenes)
        ↓
Video Recording (Kathryn screenshares + narrates) → Vimeo → embed code
        ↓
Campaign Page (timer-based, Vimeo embed, sent to list)
        ↓
GIFs of micro tools (created + hosted on Cloudinary)
        ↓
Evergreen Assembly (integrated single-scroll experience)

What Claude Code produces: Steps 1–5, 7, 9 What Kathryn does manually: Step 6 (video recording), Step 8 (GIFs)

Step 1: Discovery → Concept Brief

Triggered by: Kathryn uploads transcripts and/or JSON analysis files

What Claude reads:

• content-pipeline/discovery-sop.md — the instructions for analyzing transcripts and producing concept briefs
• Uploaded transcripts + JSON files

Process:

• Analyze transcripts for patterns across multiple clients
• Identify the expensive problem, the frequency, the current broken approach, the actual cost
• Produce a concept brief following the template in the SOP

Quality gates (from the SOP):

• Problem is specific enough to happen TODAY
• Cost has real numbers and calculation method
• Framework has named steps, not just concepts
• One toolkit direction picked (not combining)
• One workshop focus picked (not combining)
• All sections complete

Output: Concept brief → content-pipeline/concept-briefs/

The One Brief Rule: Create multiple briefs if you want. Save them. But only take ONE to production at a time. Finish and ship before starting the next.

STOP — Human Review

Kathryn reviews the concept brief and selects which one to produce. She may refine the angle, adjust the framework name, or sharpen the problem statement before proceeding.

Step 2: Interactive Article

Triggered by: Approved concept brief

What Claude reads:

• Approved concept brief from Step 1
• content/frameworks/kit-aos-interactive-narrative/ — full framework (6 files: start here, context, terminology, golden example, quality, output skill)
• OR content/frameworks/kit-standard-interactive-narrative/ (5 files) depending on brand
• skills/interactive-narrative-SKILL.md — the output skill
• brand/ — brand kit, visual style guide
• golden-examples/thought-leadership-article-golden-example-*.html — reference

Output: Article HTML → campaigns/[campaign-name]/

Quality gate: Article quality checklist (frameworks/.../04-article-quality.md)

STOP — Human Review

Kathryn reviews the article for narrative flow, accuracy, brand voice, and whether the interactive elements earn their place.

Step 3: Micro Tools (2)

Triggered by: Article is approved. Concept brief identifies which types of tools produce the most "aha."

Decide which tool types to build based on the concept:

Tool types blend. Not every tool fits neatly into one category. The Triage Call Decoder is revelation (decodes hidden meaning), diagnostic (maps to scoring criteria), and calculator (computes a composite signal) in one tool. When a concept calls for it, read multiple skill files and let the tool take the shape the "aha" requires. The categories are starting points, not containers.

What Claude reads:

• Approved concept brief
• The relevant tool skill(s) — may be more than one if the tool blends types
• brand/ — brand kit, visual style guide
• golden-examples/ — calculator, diagnostic, or revelation golden example
• The article from Step 2 (to ensure tools complement, not repeat)

Output: 2 micro tool HTMLs → campaigns/[campaign-name]/

Quality gates:

1. qc-agents/micro-tool-build-qc.md — Sections 1–10: brand/visual compliance (var() elimination, canonical colors, sharp geometry, label spec, nav/footer, typography, scrollbar, inline JS styles)
2. qc-agents/micro-tool-build-qc.md — Section 11: scoring validation (consistent personas, realistic mixed personas, tie-break verification, reachability, cross-variable coherence). Run after scoring logic is implemented and before copy QC.

STOP — Human Review

Kathryn reviews each tool. The test: does the tool make the prospect apply the framework to their own situation? If it just teaches, it's an article section, not a tool.

Step 4: Briefing Script

Triggered by: Article and micro tools approved

What Claude reads:

• Approved concept brief
• The finished article (to ensure the briefing tells the STORY, not restates the FRAMEWORK)
• skills/briefing-script-SKILL.md — the script skill (7-beat arc, stage directions, voice/visual split, timing, tone)
• skills/briefing-interactive-SKILL.md — the walkthrough HTML skill (to understand what's on screen)
• golden-examples/proof-gap-briefing-golden-example.html — reference
• Existing briefing script examples (e.g., proof-gap-briefing-script-v3.md)

The core distinction: The article teaches the framework. The briefing shows what happened when the framework was applied to a real client. If you removed the article, the briefing should still make sense as a standalone story.

Key rules (from the skill):

• Stage directions use SCROLL, HOVER, PAUSE only — no CLICK, no REVEAL, no TOGGLE
• Voice adds what the screen doesn't show — never narrate what's visible
• 7-beat arc: Practice → Discovery → Findings → Rewrite → Results → Gap → Offer
• Target runtime: 15–20 minutes
• Beat 6 (The Gap) is the conversion hinge — don't rush it

Output: Briefing script (markdown with scroll/hover/pause cues) → campaigns/[campaign-name]/

Quality gate: qc-agents/qc-checklist-briefing-script.md

STOP — Human Review

Kathryn reviews the script for narrative arc, specific client details (not generic), timing (~17 min), voice/visual split, and whether it demonstrates her diagnostic thinking. Reads it aloud while scrolling through the interactive HTML.

Step 5: Briefing Walkthrough HTML

Triggered by: Briefing script approved

What Claude reads:

• Approved briefing script from Step 4
• skills/briefing-interactive-SKILL.md — the build skill
• qc-agents/qc-checklist-briefing-interactive.md — the QC checklist
• brand/ — brand kit (Advisory OS dark theme)
• golden-examples/proof-gap-briefing-golden-example.html — reference

Key rules (from the QC):

• Zero buttons. Everything is hover or scroll.
• The presenter controls all pacing.
• 7-beat narrative arc: Practice → Discovery → Findings → Rewrite/Build → Results → Gap → Decision
• Specific client details, not framework concepts

Output: Walkthrough HTML → campaigns/[campaign-name]/

Quality gate: qc-agents/qc-checklist-briefing-interactive.md

STOP — Human Review

Kathryn reviews the walkthrough and does a dry run — scrolls through it while reading the script to check pacing and flow.

Step 6: Video Recording (Kathryn — manual)

Kathryn does this outside Claude Code:

1. Opens the walkthrough HTML in Chrome
2. Screen-records while narrating from the briefing script
3. Edits if needed
4. Uploads to Vimeo
5. Gets the embed code

Output: Vimeo embed code → give to Claude for Steps 7 and 9

Step 7: Campaign Page (Timer-Based)

Triggered by: Video is on Vimeo, embed code is ready

What Claude reads:

• Vimeo embed code
• skills/briefing-landing-page-SKILL.md — the build skill (page structure, state management, CONFIG block, content that changes per campaign, build sequence)
• golden-examples/timer-based-briefing-landing-page-golden-example.html — reference
• brand/ — brand kit
• Campaign details (when it drops, how long the timer runs, expiration date in UTC)

Key rules (from the skill):

• Two states: watch (before expiration) and expired (after expiration)
• CONFIG block with UTC expiration date drives all state logic
• Context section sets up WHY to watch — it doesn't summarize the briefing
• Expired state shows four path cards (article, tool 1, tool 2, diagnostic) — not a dead end
• Vimeo starts as about:blank placeholder, swapped after recording

Output: Timer-based campaign page HTML → campaigns/[campaign-name]/

Quality gate: qc-agents/qc-checklist-briefing-landing-page.md

STOP — Human Review

Kathryn reviews the campaign page. This gets sent to the list — final check on copy, timer, video embed, CTA, expiration date, and both watch/expired states.

Step 8: GIFs (Kathryn — manual)

Kathryn does this outside Claude Code:

1. Screen-records short demos of each micro tool in action
2. Converts to GIF
3. Uploads to Cloudinary
4. Gets the hosted URLs

Output: Cloudinary GIF URLs → give to Claude for Step 9

Step 9: Evergreen Assembly

Triggered by: All campaign pieces are approved, video is on Vimeo, GIFs are hosted

This is NOT packaging. This is recomposition. The evergreen is a single integrated scroll experience that weaves the article narrative, the micro tools (embedded inline), the briefing video, and a conversion mechanism into one continuous journey.

What Claude reads:

• The finished article from Step 2
• The finished micro tools from Step 3
• Vimeo embed code from Step 6
• GIF URLs from Step 8
• golden-examples/evergreen-framework-golden-example.html — the reference for structure
• brand/ — brand kit, visual style guide
• content/business/marketing/evergreen/proof-gap-framework.html — the Proof Gap as the primary reference

The structure (from the Proof Gap reference):

1. Hero + section nav
2. Article narrative sections (rewritten to flow through the full journey)
3. Micro tools embedded inline as interactive sections (with GIF previews)
4. Briefing video section (Vimeo embed)
5. "The Gap You Can't Close Alone" — conversion mechanism
6. CTA — Book a Diagnostic

Key principle: Someone scrolls through the entire experience without clicking away. The article makes them aware. The tools make them diagnose. The video proves the expertise. The conversion section makes the CTA obvious. One scroll, one journey.

Output: Evergreen HTML → evergreen/

Quality gate: qc-agents/copy-qc.md + brand QC

STOP — Human Review

Kathryn reviews the evergreen thoroughly. This is the asset that works forever. Extra care on narrative flow, tool integration, mobile responsiveness, and whether the scroll creates the journey from "I didn't know I had this problem" to "I need to book a call."

After Production — Push Changes

In Claude Code: "Push the vault" — Claude commits and pushes all new campaign and evergreen files to GitHub.

Folder Structure

content/business/marketing/
├── content-pipeline/
│   ├── discovery-sop.md                    ← the instructions
│   └── concept-briefs/                     ← the sparks
│       ├── 01-pain-dissipation-gap.md
│       └── ...
├── campaigns/
│   ├── proof-gap/                          ← all pieces for one campaign
│   │   ├── advisory-os-proof-gap.html      ← article
│   │   ├── proof-gap-story-finder.html     ← micro tool 1
│   │   ├── proof-gap-readiness-check.html  ← micro tool 2
│   │   ├── proof-gap-briefing-script.md    ← script
│   │   ├── proof-gap-walkthrough.html      ← recording visual
│   │   └── proof-gap-briefing.html         ← timer campaign page
│   ├── subtract-add/
│   │   └── ...
│   └── [next-campaign]/
├── evergreen/
│   ├── proof-gap-framework.html
│   ├── subtract-add-framework.html
│   └── [next-evergreen].html
├── golden-examples/                        ← reference only, don't modify
├── skills/
├── qc-agents/
├── brand/
└── CONTENT-WORKFLOW.md                     ← this file

Skills Quick Reference

QC Quick Reference