Home Use Cases Work Insights About Contact
Back to Our Work

How a Growth-Stage Marketing Agency Built Keyword Gap Articles with Human Approval Before Budget Is Committed

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. Step 3 is waiting for approval right now. Click it 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 Marketing Agency Built Keyword Gap Articles with Human Approval Before Budget Is Committed

Before
A person used to write every piece from a blank page.
A person used to load, format and schedule every piece by hand.
A person used to pick every headline by feel.
After
the month's topics, argued from the data before anybody writes
every piece written to its brief, in the client's voice
the short queue an editor clears before anything ships
every cleared piece live where and when it should be
two versions fight, the numbers pick, nobody guesses
Executive Summary

Turns a client's keyword gaps into published pieces that sound like the client wrote them, and never lets a claim without a source go out under a client's name. Written for an agency where the record is the point: every brief carries the data that argued for it, every published value traces back to its draft, and the only things left for a person are the calendar, the clearance and the signature on the month. Six stations, run as a loop: what station 6 learns about August is what station 1 briefs for September.

What Was Broken

A person used to write every piece from a blank page
A person used to load, format and schedule every piece by hand
The real cost
A person used to pick every headline by feel.

What We Built

Six stations and 12 subagents. Each subagent carries its own tasks and its own refusal.

A4
Keyword Research
Track rankings for every keyword the client competes on, week over week
A1
Brief Drafting
Draft a brief per topic: the angle, the keywords, the structure, the sources
A1
Draft Writing
Write each piece to its brief, in the client's voice, at the agreed length
A3
Voice Check
Check every draft against the client's voice guide, line by line
A3
Fact Check
Check every claim in the draft against the source the brief named
A2
Review Brief
Summarize each draft into the few lines the editor decides from
A3
Channel Formatting
Convert each cleared piece into each channel's format: post, email, social card
A4
Publish Scheduling
Slot each piece into its approved calendar date, per channel
A4
Variant Serving
Serve the approved title and meta variants in rotation
A3
Winner Calling
Compute whether the gap between variants is real or noise
A4
Traffic Attribution
Pull traffic, rankings and conversions for every live piece
A1
Report Drafting
Draft the month's report: what ran, what moved, what it brought in

How It Runs

1

Topic Planning

A person answers here

Agents track every keyword the client competes on, find the gaps competitors have not filled, and draft a brief for each topic before anybody writes a word. A person approves the calendar, because the plan is where the client's budget gets committed.

2

Draft Writing

Nobody is asked anything here

Agents write each piece to its brief in the client's voice, weave the keywords in as sentences, and draft the title and meta variants the test will later need. Nobody is asked anything here. A draft that stops on a missing source, or drifts off voice, reaches the editor at station 3 with the lines already marked.

3

Editorial Review

A person answers here

Agents check every claim against its source and assemble each draft into a card the editor can clear from a few lines, with the fix for every flag already drafted. This is the station where a person works. Everything before it exists to make this read short, and nothing publishes without it.

4

Channel Publishing

Nobody is asked anything here

Agents convert every cleared piece into each channel's own format, publish on schedule, and confirm each one is live and indexed where it should be. Nobody presses publish any more. A piece that breaks in conversion holds itself, and reaches the editor with the broken element named.

5

Headline Testing

Nobody is asked anything here

Agents run the editor-approved title and meta variants against each other, split the traffic evenly, and lock the winner only when the gap between them is real. Nobody picks a headline by feel any more. The variants were approved at station 3; here the numbers decide between them.

6

Performance Reporting

A person answers here

Agents trace every lead back to the piece and the keyword that produced it, draft the month's report, and map what ranked and converted into next month's candidate topics. A person signs the report before the client sees it. That signature stays human, because somebody must be answerable for numbers a client pays against.

Where a Person Decides

Step 1, Topic Planning. Approve the calendar, or strike the topics that do not belong on it. Read the flagged keyword and decide who it is actually for. the plan commits the client's budget.
Step 3, Editorial Review. Read each card and clear it, return it, or strike the claim. Open the full draft only when the card is not enough. the client's name is on it.
Step 6, Performance Reporting. Settle the unexplained Tuesday drop, or send it to the client as open. Accept or strike the refresh list, then sign the report. somebody answers for the numbers.

Operating Model

This changes how work flows through the team.

Role
Responsibility
Topic Planning owner
Approve the calendar, or strike the topics that do not belong on it. Read the flagged keyword and decide who it is actually for. the plan commits the client's budget.
Editorial Review owner
Read each card and clear it, return it, or strike the claim. Open the full draft only when the card is not enough. the client's name is on it.
Performance Reporting owner
Settle the unexplained Tuesday drop, or send it to the client as open. Accept or strike the refresh list, then sign the report. somebody answers for the numbers.
Channel Publishing, when it goes wrong
Look at the calendar if you want to see what lands this week.

What Transfers, What Must Be True

What transfers
A person is in the loop wherever the client's budget is committed, the client's name goes out, or a number reaches the client. Three of the six stations refuse to proceed on their own, Topic Planning, Editorial Review and Performance Reporting, 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 Planning stops for a person, and the plan commits the client's budget.
Editorial Review stops for a person, and the client's name is on it.
Performance Reporting stops for a person, and somebody answers for the numbers.
Every subagent says what it will not do. 12 of them do.
What must be true in your environment
The agent can read the systems your records already live in. This one reads 18.
Somebody owns Topic Planning and has time for it.
Somebody owns Editorial Review and has time for it.
Somebody owns Performance Reporting and has time for it.

Failure Modes

What breaks this pattern:

✗ Unsourced claims ship as facts

The writer fills a source gap with a sentence that sounds right. The client publishes a claim nobody can defend when a reader asks where it came from.

✗ New pages cannibalize old ones

The calendar schedules a topic the site already ranks for. The new page and the old page split the search traffic, and both rank lower than the old page did alone.

✗ Headline winners called on noise

The test picks a winner while the gap between variants is still random. Every headline after that gets shaped by a result that was never real.

✗ Leads credited to the wrong page

A lead the system cannot trace gets credited to the nearest page. The client pays for more of the content that never produced a lead.

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.

Drafts written
Counted at Draft Writing
74
Claims invented to fill a gap
Counted at Draft Writing
0
Drafts that stopped for a source
Counted at Draft Writing
3
Pieces published
Counted at Channel Publishing
70
Published broken
Counted at Channel Publishing
0
Held and handed to a person
Counted at Channel Publishing
1
Tests run
Counted at Headline Testing
28
Winners called on noise
Counted at Headline Testing
0
Tests it refused to call
Counted at Headline Testing
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 Planning
GATE
subagenttopic-research
writesplan/briefs/<slug>.json
raisesapprove-calendar
may touchplan/**, reports/** read-only
2Draft Writing
subagentdraft-writer
writesdrafts/<slug>.md
raisesclaim-without-source
may touchdrafts/**, plan/briefs/** read-only, voice/** read-only
3Editorial Review
GATE
subagentreview-brief
writesreview/<slug>.json
raisesawait-editor-clearance
may touchreview/**, drafts/** read-only
4Channel Publishing
subagentpublish-run
writespublished/<slug>.json
raisesbroken-render
may touchpublished/**, review/** read-only
5Headline Testing
subagentvariant-test
writestests/<slug>.json
raisestest-refused
may touchtests/**, published/** read-only
6Performance Reporting
GATE
subagentmonth-report
writesreports/<period>.json
raisessign-report
may touchreports/**, tests/** read-only, plan/** read-only

Stack

Every system this agent reads or writes.

System
Read at
Stations
GA4
Performance Reporting
1 of 6
Google Search Console
Topic Planning
3 of 6
the LinkedIn scheduler
Channel Publishing
1 of 6
the WordPress site
Channel Publishing
2 of 6
the analytics feed
Headline Testing
1 of 6
the brief each draft was written to
Editorial Review
1 of 6
the brief library
Draft Writing
1 of 6
the client's CRM
Performance Reporting
1 of 6
the client's published archive
Topic Planning
2 of 6
the client's voice guide
Draft Writing
2 of 6
the content calendar
Topic Planning
1 of 6
the email platform
Channel Publishing
2 of 6
the monthly report
Performance Reporting
1 of 6
the rank tracker
Topic Planning
1 of 6
the review queue
Editorial Review
1 of 6
the source library
Editorial Review
1 of 6
the test log
Headline Testing
1 of 6
the writing models
Draft Writing
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