Home Use Cases Work Insights About Contact
Back to Our Work

How a Growth-Stage HR Services Company Built Content Publishing Where Every Draft Carries Its Sources

Marketing1
Open the walkthrough

Six-Step AI Workflow with Human Oversight

Each step shows the workspace, what the AI completed, and what still needs human review.

Steps 1, 3 and 6 need a person. Steps 1 and 3 are waiting for approval right now. Click either to open the full walkthrough.

Estimates are directional and based on stated assumptions. All names, organizations, and identifying details have been anonymized in accordance with our confidentiality agreements.

The Transformation

How a Growth-Stage HR Services Company Built Content Publishing Where Every Draft Carries Its Sources

Before
A writer used to type each draft out from the brief and read old posts to keep the voice.
A coordinator used to keep the calendar in a spreadsheet and paste each newsletter and social version by hand.
An analyst used to pull the numbers into a spreadsheet each Monday and guess which page earned each lead.
After
every topic scored, and a brief for each one worth chasing
the brief written out in full, with every claim carrying its source
every claim opened against its source, and one card per draft for the editor
every cleared draft out on its slot, checked live before its versions follow
what every live piece earned, with the trace behind every number
Executive Summary

Keeps the publishing calendar fed without letting the brand voice thin, and never lets a claim it could not source go out under the company's name. Written for a marketing team that publishes more than it can hand-check: every draft carries its sources, every number on the dashboard carries its trace, and the person's work shrinks to picking topics, clearing drafts and signing the quarterly review. Six stations, run as a loop: what station 6 learns re-aims what station 1 briefs.

What Was Broken

A writer used to type each draft out from the brief and read old posts to keep the voice
A coordinator used to keep the calendar in a spreadsheet and paste each newsletter and social version by hand
The real cost
An analyst used to pull the numbers into a spreadsheet each Monday and guess which page earned each lead.

What We Built

Six stations and 14 subagents. Each subagent carries its own tasks and its own refusal.

A3
Keyword Research
Track every query the site ranks for, and where each one moved this week
A3
Competitor Watch
Scan the pages that outrank the site for each target term
A1
Brief Drafting
Draft a brief per topic: the term it chases, the intent, the outline, the internal links
A1
Copy Drafting
Draft the article section by section against the brief's outline
A2
Voice Matching
Compare each draft against the voice samples the company signed off
A3
Page Structuring
Check the title, description, headings and links against the brief's term
A3
Claim Checking
Check every factual claim in the draft against the source it cites
A1
Review Briefing
Summarize what the draft claims, chases and links, on one card
A4
Calendar Scheduling
Slot each cleared draft into the calendar against the cadence each channel holds
A2
Version Cutting
Cut each channel's version from the live page: newsletter block, social post, snippet
A4
Result Tracing
Pull rankings, traffic and time-on-page for every published piece, daily
A2
Variant Testing
Draft two variants of the title and the call to action for a page underperforming its position
A3
Refresh Triage
Sort the library into keep, refresh and retire, with the evidence for each call
A1
Quarter Reporting
Draft the quarter's report: what published, what it earned, what it taught

How It Runs

1

Topic Selection

A person answers here

Agents track what the site ranks for, read what outranks it, and draft a brief for each topic worth a calendar slot. The person picks which briefs earn a slot. The calendar is a promise the team makes, so an agent does not make it.

2

Draft Writing

Nobody is asked anything here

Agents write each committed brief into a full draft, pull it into the house voice, and validate it against the article schema before it can move. Nothing waits on a person here. A draft reaches somebody at station 3, and only after every gap in it is closed.

3

Editorial Review

A person answers here

Agents open every source the draft cites, route the legal-adjacent passages to the compliance reader, and put each waiting draft on one card the editor can decide from. This is where a person works on every piece. The byline is a person's word, and an agent cannot give it.

4

Channel Publishing

Nobody is asked anything here

Agents slot each cleared draft into the calendar, cut the newsletter and social versions from the live page, and check every version against the page it came from before it goes out. A person signs nothing here. The signature already happened at station 3, and a failed live check pulls a version back on its own.

5

Performance Tracking

Nobody is asked anything here

Agents pull rankings, traffic and conversions for every live piece, trace each lead back to the pages that carried it, and test title and call-to-action variants where a page underperforms its position. Nothing waits on a person here. The week's numbers land in Monday's digest, and only traced numbers land at all.

6

Library Review

A person answers here

Agents sort the whole library into keep, refresh and retire with the evidence for each call, queue the refresh briefs back into station 1, and draft the quarter's report. A person signs the quarter. Retiring a page and redirecting its address is hard to reverse once other sites link to it, so that call stays human.

Where a Person Decides

Step 1, Topic Selection. Read the 4 briefs and pick which go on the calendar. Decide whether the pricing topic waits for a pricing page or gets dropped. the calendar is a promise you make.
Step 3, Editorial Review. Read two cards and clear, edit or return each draft. Chase the compliance passage only if it passes its second day. the byline is a person's word.
Step 6, Library Review. Decide the 4 retirals; a redirect is hard to reverse once other sites link to it. Sign the quarter, and next quarter's slots follow your signature. retired pages do not come back.

Operating Model

This changes how work flows through the team.

Role
Responsibility
Topic Selection owner
Read the 4 briefs and pick which go on the calendar. Decide whether the pricing topic waits for a pricing page or gets dropped. the calendar is a promise you make.
Editorial Review owner
Read two cards and clear, edit or return each draft. Chase the compliance passage only if it passes its second day. the byline is a person's word.
Library Review owner
Decide the 4 retirals; a redirect is hard to reverse once other sites link to it. Sign the quarter, and next quarter's slots follow your signature. retired pages do not come back.

What Transfers, What Must Be True

What transfers
A person is in the loop wherever the company's name goes on record or a page comes down for good. Three of the six stations refuse to proceed on their own, Topic Selection, Editorial Review and Library Review, and those are the three a person carries. Everywhere else the agents run at volume and reach you only when they cannot settle something.
Topic Selection stops for a person, and the calendar is a promise you make.
Editorial Review stops for a person, and the byline is a person's word.
Library Review stops for a person, and retired pages do not come back.
Every subagent says what it will not do. 14 of them do.
What must be true in your environment
The agent can read the systems your records already live in. This one reads 20.
Somebody owns Topic Selection and has time for it.
Somebody owns Editorial Review and has time for it.
Somebody owns Library Review and has time for it.

Failure Modes

What breaks this pattern:

✗ Search volume nobody measured

The agent cannot verify a term's volume, so it carries a number anyway. Every topic decision downstream now rests on demand nobody measured.

✗ Facts invented to fill gaps

The brief calls for a stat and no source holds it, so the draft makes one up. The invented fact reads exactly like a real one. It ships under the company's name.

✗ Claims pass without a source

The reviewer cannot open the source and passes the claim anyway. The first person to actually check it is a customer.

✗ Leads credited to the nearest page

A lead with a broken path gets handed to whatever page sits closest. The report says that page works when nobody knows if it does, and next quarter's budget follows the report.

Directional Outcomes

What the agent counts, and the station that counts it.

These counts are the tallies from one monitored run of the agents. They are not monthly or annual totals.

sections drafted against the brief's outline
Counted at Draft Writing
7
facts invented where no source held
Counted at Draft Writing
0
marked gaps sent back for a source
Counted at Draft Writing
1
versions cut and queued this week
Counted at Channel Publishing
9
versions shipped past a failed live check
Counted at Channel Publishing
0
pulled cuts reported with the failed check
Counted at Channel Publishing
1
form fills and bookings traced whole
Counted at Performance Tracking
16
leads credited to the nearest page
Counted at Performance Tracking
0
bookings held as unattributed in the digest
Counted at Performance Tracking
2
Our measurement policy: We do not publish precise ROI without baseline methodology. Every figure above carries its basis.

What Runs Where

Every step names the subagent that does the work, the record it writes, the thing that raises a question for a person, and what it is allowed to touch. This is drawn from the source, not from a diagram somebody kept in sync by hand.

1Topic Selection
GATE
subagenttopic-scout
writespipeline/briefs/<brief-id>.json
raisescalendar-commit
may touchpipeline/**, library/** read-only
2Draft Writing
subagentdraft-writer
writespipeline/drafts/<draft-id>.md
raisesunsupported-claim
may touchpipeline/**, voice/** read-only
3Editorial Review
GATE
subagentreview-brief
writespipeline/review/<draft-id>.json
raisesawait-editor-decision
may touchpipeline/review/**, everything else read-only
4Channel Publishing
subagentpublish-run
writespublished/<slug>/manifest.json
raiseslive-check-failed
may touchpublished/**, pipeline/review/** read-only
5Performance Tracking
subagentperf-reader
writesanalytics/weekly/<week>.json
raisestest-inconclusive
may touchanalytics/**, published/** read-only
6Library Review
GATE
subagentlibrary-review
writeslibrary/review-<quarter>.json
raisessign-quarter-review
may touchlibrary/**, analytics/** read-only

Stack

Every system this agent reads or writes.

System
Read at
Stations
the CMS
Channel Publishing
1 of 6
the CRM
Performance Tracking
1 of 6
the analytics property
Performance Tracking
1 of 6
the article schema
Draft Writing
1 of 6
the brief queue
Draft Writing
1 of 6
the competitor watch list
Topic Selection
1 of 6
the compliance reader's queue
Editorial Review
1 of 6
the editorial calendar
Topic Selection
3 of 6
the newsletter platform
Channel Publishing
1 of 6
the publish notes
Editorial Review
1 of 6
the published library
Library Review
1 of 6
the quarterly report
Library Review
1 of 6
the rank tracker
Topic Selection
2 of 6
the redirect map
Library Review
1 of 6
the review queue
Editorial Review
1 of 6
the search console
Topic Selection
1 of 6
the signed-off voice samples
Draft Writing
1 of 6
the social scheduler
Channel Publishing
1 of 6
the source library
Draft Writing
2 of 6
the testing tool
Performance Tracking
1 of 6
Next Step

Want to see if this pattern fits your marketing?

No build commitment·Real samples, not a demo·Estimate in writing