Home Use Cases Work Insights About Contact
Back to Our Work

How a Growth-Stage Educational Institution Built Articles That Cite the Course Catalogue Behind Every Claim

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 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 Educational Institution Built Articles That Cite the Course Catalogue Behind Every Claim

Before
A person used to spend the first week of every month picking topics by hand.
A person used to write every post from a blank page.
A person used to paste each post into the site, the newsletter and each social account by hand.
A person used to pick between two headlines by feel.
After
every topic scanned, the ones worth writing briefed
every brief written into a cited, tuned draft
the short queue of drafts waiting on the editor
the calendar filled, pieces live on every channel
two versions compete, the numbers pick one
Executive Summary

Keeps a growth-stage school publishing faster than its team can write, and never lets a claim about a programme go out that the course catalogue does not back. Written for an org where the record is the point: every draft cites the catalogue line behind every fact, every headline winner is picked by the numbers, and the only things left for a person are the drafts waiting on the editor and the month waiting on a signature. Six stations, run as a loop: what station 6 counts decides what station 1 briefs.

What Was Broken

A person used to spend the first week of every month picking topics by hand
A person used to write every post from a blank page
A person used to paste each post into the site, the newsletter and each social account by hand
The real cost
A person used to pick between two headlines by feel.

What We Built

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

A4
Search Scout
Track the search terms prospective students use around each programme
A1
Brief Writer
Draft a brief per topic: the search term, the reader, the angle, the pages to beat
A2
Copy Writer
Write each brief into a full draft: the opening, the sections, the questions readers ask
A3
Voice Check
Check every draft against the school's voice guide, section by section
A2
Page Tuning
Set the title, description and headings around the search term the brief names
A2
Review Pack
Assemble each draft for review: the brief beside it, every flag, every cited fact
A4
Queue Keeper
Order the review queue by each draft's slot on the publishing calendar
A4
Calendar Slotting
Slot each cleared piece into the publishing calendar
A3
Channel Push
Format each piece for the channel it goes to: site page, newsletter cut, social post
A1
Variant Writer
Draft two headline-and-opening variants for each page picked for a test
A4
Test Runner
Split each page's visitors evenly across its two variants
A4
Rank Watch
Track where every published page ranks for the term its brief named
A3
Lead Ledger
Trace each enquiry and application back to the pages the person read
A1
Plan Drafter
Draft next month's brief list from what ranked, what converted and what the tests taught

How It Runs

1

Topic Research

Nobody is asked anything here

Agents scan what prospective students actually search for, watch the pages competitors already hold, and write a brief for every topic worth the school's time. Nothing waits on a person here. The plan signed at station 6 is what fills this queue.

2

Drafting

Nobody is asked anything here

Agents write each brief into a full draft in the school's voice, cite the catalogue line behind every programme fact, and set the title, headings and links around the term the brief named. A person sees the draft only at station 3, with every flag already attached to it.

3

Editorial Review

A person answers here

Agents assemble each draft for the editor: the brief beside it, every flag, every cited fact one click deep, and the three lines that say what needs ruling on. This is where a person works. Nothing carries the school's name until the editor says so.

4

Publishing

Nobody is asked anything here

Agents slot each cleared piece into the calendar, format it for the site, the newsletter and the social accounts, push it live, and check the live page before any reader finds it broken. Nobody signs here. The editor's ruling at station 3 was the last signature a piece needs.

5

Headline Testing

Nobody is asked anything here

Agents write two openings for a page picked for a test, split its visitors between them, count enquiries rather than clicks, and roll out the winner when the numbers can carry the call. Nobody picks a winner here. The numbers do, and a person reads the closed tests when they want to know what framing wins.

6

Performance Review

A person answers here

Agents track every published page against the term its brief named, trace each enquiry back to the pages that earned it, and draft next month's brief list from what the month taught. A person signs the month and the next plan, because somebody must own the numbers.

Where a Person Decides

Step 3, Editorial Review. Read the three lines and rule on each draft. Open the full draft only when the lines are not enough. the school's name is on every page.
Step 6, Performance Review. Settle the nursing drop: accept the flag or name the cause. Accept or trim the November briefs, then sign the plan. somebody must own the numbers.

Operating Model

This changes how work flows through the team.

Role
Responsibility
Editorial Review owner
Read the three lines and rule on each draft. Open the full draft only when the lines are not enough. the school's name is on every page.
Performance Review owner
Settle the nursing drop: accept the flag or name the cause. Accept or trim the November briefs, then sign the plan. somebody must own the numbers.

What Transfers, What Must Be True

What transfers
A person is in the loop wherever the school's name goes on a page or a month's numbers get owned. Two of the six stations refuse to proceed on their own, Editorial Review and Performance Review, and those are the two a person carries. Everywhere else the agents run at volume and reach you only when they cannot settle something.
Editorial Review stops for a person, and the school's name is on every page.
Performance Review stops for a person, and somebody must own the numbers.
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 18.
Somebody owns Editorial Review and has time for it.
Somebody owns Performance Review and has time for it.

Failure Modes

What breaks this pattern:

✗ Guessed demand fills the pipeline

Nobody sources the demand number, so the brief carries a guess. The school pays a writer for pages on topics nobody searches for.

✗ Drafts go out without sign-off

A draft goes live under the school's name before the editor says so. The first person to catch a wrong fee is a parent.

✗ A winner declared by feel

The test ends when someone likes a headline, not when the numbers settle. The team then copies that pattern to every other page.

✗ Every enquiry gets credited anyway

An enquiry the system cannot trace gets assigned to a likely page. The monthly report then rewards pages that never earned a lead, and the budget follows them.

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.

Topics scanned
Counted at Topic Research
120
Briefs written on a guessed demand number
Counted at Topic Research
0
Held for the monthly review
Counted at Topic Research
1
Drafts written
Counted at Drafting
34
Programme facts stated without a catalogue line
Counted at Drafting
0
Questions handed to the editor
Counted at Drafting
7
Pieces pushed live
Counted at Publishing
30
Broken pages left up
Counted at Publishing
0
Pulled and handed back
Counted at Publishing
1
Tests run
Counted at Headline Testing
12
Winners called before the numbers carried it
Counted at Headline Testing
0
Draws and declines handed to the monthly review
Counted at Headline Testing
5
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 Research
subagenttopic-scout
writesbriefs/<topic>.json
raisesdemand-unsourced
may touchbriefs/**, rankings/** read-only
2Drafting
subagentwrite-draft
writesdrafts/<slug>.md
raisesfact-not-in-catalogue
may touchdrafts/**, briefs/** and catalogue/** read-only
3Editorial Review
GATE
subagentreview-pack
writesreview/<slug>.json
raisesawait-editor-ruling
may touchreview/**, everything else read-only
4Publishing
subagentpublish-channels
writespublished/<slug>.json
raiseslive-page-broken
may touchpublished/**, review/** read-only
5Headline Testing
subagentrun-test
writestests/<slug>.json
raisestest-will-not-settle
may touchtests/**, published/** read-only
6Performance Review
GATE
subagentmonth-review
writesreports/month-<period>.json, briefs/next-month.json
raisessign-monthly-plan
may touchreports/**, tests/** and published/** read-only

Stack

Every system this agent reads or writes.

System
Read at
Stations
Google Search Console
Topic Research
2 of 6
both social accounts
Publishing
1 of 6
the Mailchimp newsletter
Publishing
1 of 6
the Registrar's office for the questions only it can settle
Editorial Review
1 of 6
the WordPress site
Publishing
2 of 6
the analytics account
Headline Testing
2 of 6
the competitor pages we watch
Topic Research
1 of 6
the course catalogue
Topic Research
3 of 6
the drafts folder in the CMS
Drafting
1 of 6
the enquiry CRM
Performance Review
1 of 6
the enquiry forms
Headline Testing
1 of 6
the image library
Drafting
1 of 6
the keyword tool
Topic Research
1 of 6
the monthly report
Performance Review
1 of 6
the publishing calendar
Editorial Review
2 of 6
the review queue in the CMS
Editorial Review
1 of 6
the school's voice guide
Drafting
1 of 6
the test log
Headline Testing
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