Every page here is also a document. Use your browser's print or save-as-PDF and you get the same PDF that goes out by email, which is why these are laid out like sheets of paper.
EZO Content Pipeline · Pillar 1 of 5 · Structured Brief · V0 walkthrough · 2026-07-28
Your brief process, mechanized - you stay the approver
The PMM Page Brief Process v1.0 (five phases, unchanged) becomes an assembly line. The machine gathers and drafts; a PMM reviews every brief before it reaches Content. Not automated - every step visible, every input inspectable.
● live / built today
◕ in motion this week
○ proposed / queued
■ blocked - named owner
NEW = capability added beyond today's process
BLUE BOX = the one human review
Sources in - what feeds the brief (each item marked with its status)
Product truth
●Fact store: 137 verified product facts, each carrying its source (137 = count of seeded facts, review pending)
◕Redmine product data mined; wiring into the input pack
■Product roadmap docs - waiting on team share
SEO inputs
●Keyword + gap sheets (SEO Manager's plan, 15 sheets = file count in the Q3 plan folder)
◕Gumshoe API feed - key in hand; replaces manual sheet updates
○SERP scrape + synthesis, top 7 (= first 7 search results per keyword) NEW
○SEMrush signals - API key requested NEW
Strategy canon (HQ is the master)
◕Marketing HQ KB: ICPs, wedge, brand voice, pricing - always cited, never overridden; ingest this week
○Website strategy outline - which pages exist and why
Process canon
●PMM Page Brief Process v1.0 - the 5 phases below, followed as written
●Gold-standard brief (Hardware Discovery) - the benchmark every machine brief is diffed against, structure only
Everything above lands in one inspectable input pack. Rule: the machine may use ONLY these sources - where one is missing it writes [NEEDS-SOURCE], it never invents.
The five phases, mechanized - then the human review
Phase 1
Load internal sources
◕Machine assembles the input pack before any external research
PMM spot-checks the sources
→
Phase 2
Page structure + template
●Machine applies the approved page-type template, with character counts
PMM confirms page type fits
→
Phase 3
Keywords from SEO
●Today: keyword sheets
◕Next: live Gumshoe feed
SEO Manager confirms before final
→
Phase 4
Gather content
●Machine drafts pains, features, value props, proof points from provided facts only; gaps become [NEEDS-SOURCE]
Nothing invented, by rule
→
Phase 5
Assemble + QA
●Machine compiles the brief + a diff report vs the gold standard
○QA checklist as automatic pre-checks
Diff report shown, not hidden
→
The human reviewer
PMM review + approve
The PMM runs the v1.0 QA checklist on the machine brief: sources cited, ICP fit, product-line integrity, keywords confirmed.
Approve, or every miss becomes a named gap with an owner. Nothing reaches Content without this sign-off.
→
Then
Pillar 2: AI draft
The approved brief is the only input the drafter gets - same grounding rule, then the QA check and one human review.
Today's session - six steps in order (circled numbers = step order, nothing else)
1Frame
The v1.0 process doc is the spec; today is V0 - your job is to beat it up.
2Open the inputs
The real fact, keyword, and template files, on screen - plain text, inspectable.
3Fire it live
One command, one model call, watched together. If it fails, it fails out loud.
4Read the output
Machine brief beside the gold brief, plus the diff report line by line.
5Your review - the check
You run the v1.0 QA checklist; every miss is written down as a named gap.
6Agree next wiring
Which sources close which gaps; you see V1 this week. Your homework: none.
Where V0 stands tonight - honestly
●Generator built: one command, one visible model call, output in plain text
●Diff report built: machine brief scored against the gold brief, gaps named
■Live run blocked: API credits not yet reaching our key - owner Brent, fix in motion
◕This session: you mark what is wrong or missing - that list is the V1 build order