Home Use Cases Work Insights About Contact
Back to Our Work

How a Growth-Stage Technology Company Built a Posting Loop with Human Sign-off Wherever the Brand Speaks

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, 5 and 6 need a person. Steps 3 and 5 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 Technology Company Built a Posting Loop with Human Sign-off Wherever the Brand Speaks

Before
A person used to plan every post in a spreadsheet on Monday morning.
A person used to write every post for every platform by hand.
A person used to post each one by hand, when they remembered, from whichever device was nearest.
After
the week's slots, drafted from what is actually moving
every variant written, cut to its platform, checked against the voice
the short queue of posts waiting for a person to lend the brand's name
every cleared post live at its slot, confirmed against the cleared copy
every conversation answered, moderated, or handed to a person with its history
Executive Summary

Keeps the brand talking on every network at the pace the platforms demand, and never lets an unsourced claim, an expired usage right, or an angry customer pass without a person. Written for a marketing team where one person used to write, post, watch and count everything by hand, and the feed went quiet whenever that person was busy. Six stations, run as a loop: what station 6 learns about the month is what station 1 plans the next one from.

What Was Broken

A person used to plan every post in a spreadsheet on Monday morning
A person used to write every post for every platform by hand
The real cost
A person used to post each one by hand, when they remembered, from whichever device was nearest.

What We Built

Six stations and 15 subagents. Each subagent carries its own tasks and its own refusal.

A4
Trend Watch
Monitor the four networks for conversations the brand could join
A1
Calendar Drafting
Draft the week's slots per platform: topic, format, day, time
A1
Post Writer
Draft the master version of each post from its slot brief and its campaign
A1
Platform Fitter
Recast each master draft to each platform's length, format and register
A3
Voice Check
Check every variant against the voice guide, line by line
A2
Review Prep
Assemble each post into one card: variants, assets, and the source of every claim
A3
Rights Check
Check every asset for usage rights: creator clips, music, stock
A4
Scheduler
Queue each cleared variant against its slot, per network
A3
Publish Guard
Confirm each post is live within a minute of its slot
A3
Reply Handling
Answer routine comments and DMs from the approved answer bank
A4
Moderation
Classify every comment as routine, spam, or needs-a-person
A3
Sentiment Watch
Score the mood of every thread as it moves
A4
Metrics Roll-up
Collect the month's numbers from each network's own reporting
A2
Insight Drafting
Draft what worked and what did not, per platform and per campaign
A1
Rule Proposal
Propose the calendar, cadence and voice-guide changes the month argues for

How It Runs

1

Calendar Planning

Nobody is asked anything here

Agents watch what is moving on each network and draft the week's slots: topic, platform, format, day and time. Nothing waits on a person here. A slot only becomes a post, and every post stops at station 3 before it publishes.

2

Post Drafting

Nobody is asked anything here

Agents write one master draft per slot, cut it to each platform's length and register, and check every line against the voice guide before it moves. Nothing waits on a person here. Every variant stops at station 3 before it can publish.

3

Brand Sign-off

A person answers here

Agents assemble each post into one card a reviewer can decide from: every variant, every asset, the source behind every claim, the rights behind every clip. This is the station where a person works. Nothing publishes under the brand's name until somebody clears it.

4

Publishing

Nobody is asked anything here

Agents queue each cleared variant against its slot, publish it on its network at its time, and confirm the live copy matches the cleared copy. Nothing waits on a person here. A failed publish retries itself, and a failing network holds its own queue.

5

Community Engagement

A person answers here

Agents answer the routine questions from the approved answer bank, hide what the posted rules name, and escalate a thread that is heating up with its whole history attached. This is a station where a person works. An escalated thread gets no automated reply until somebody takes it.

6

Month Review

A person answers here

Agents pull the month's numbers from every network, reconcile them against what was actually published, trace what moved back to the posts that moved it, and propose the changes next month should run on. This is a station where a person works. No rule the agents run on changes until somebody accepts it.

Where a Person Decides

Step 3, Brand Sign-off. Read each card and clear it, or send it back with one line. Open the full post only when the card is not enough. the brand's name is yours to lend.
Step 5, Community Engagement. Take the 2 escalated threads; the history is already attached to each. Add your answer to the bank if the question will come back. an angry customer gets a person.
Step 6, Month Review. Accept or reject the 2 proposed changes, and call the creator shortlist. Sign the month, or leave the unexplained dip open with a note. next month's rules are yours to set.

Operating Model

This changes how work flows through the team.

Role
Responsibility
Brand Sign-off owner
Read each card and clear it, or send it back with one line. Open the full post only when the card is not enough. the brand's name is yours to lend.
Community Engagement owner
Take the 2 escalated threads; the history is already attached to each. Add your answer to the bank if the question will come back. an angry customer gets a person.
Month Review owner
Accept or reject the 2 proposed changes, and call the creator shortlist. Sign the month, or leave the unexplained dip open with a note. next month's rules are yours to set.

What Transfers, What Must Be True

What transfers
A person is in the loop wherever the brand speaks or a rule changes. Three of the six stations refuse to proceed on their own, Brand Sign-off, Community Engagement and Month 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.
Brand Sign-off stops for a person, and the brand's name is yours to lend.
Community Engagement stops for a person, and an angry customer gets a person.
Month Review stops for a person, and next month's rules are yours to set.
Every subagent says what it will not do. 15 of them do.
What must be true in your environment
The agent can read the systems your records already live in. This one reads 26.
Somebody owns Brand Sign-off and has time for it.
Somebody owns Community Engagement and has time for it.
Somebody owns Month Review and has time for it.

Failure Modes

What breaks this pattern:

✗ Drafts built on invented sources

A brief arrives with no source and the writer fills the gap with a plausible fact. The post reads fine, and the claim behind it does not exist.

✗ Posts running on expired rights

A photo license runs out and the post stays up. The rights holder finds it before you do.

✗ The queue fires into a failing network

A network's API starts failing and the scheduler keeps sending into it. Some posts go out twice, some are lost, and you cannot tell which.

✗ Replies improvised in your name

A customer asks a question the playbook does not cover. The agent writes its best guess and answers as the brand.

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.

Slots planned
Counted at Calendar Planning
124
Published without sign-off
Counted at Calendar Planning
0
Topics held for a person's call
Counted at Calendar Planning
3
Variants drafted
Counted at Post Drafting
356
Claims invented
Counted at Post Drafting
0
Claims flagged for sign-off
Counted at Post Drafting
12
Variants published
Counted at Publishing
349
Went out unsigned or off its cleared copy
Counted at Publishing
0
Held for a network failure
Counted at Publishing
3
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.

1Calendar Planning
subagentcalendar-draft
writescalendar/<week>/slots.json
raisestopic-outside-brand
may touchcalendar/**, insights/** read-only
2Post Drafting
subagentpost-draft
writesposts/<post-id>/drafts/
raisesclaim-without-source
may touchposts/**, assets/** read-only, calendar/** read-only
3Brand Sign-off
GATE
subagentsignoff-prep
writesposts/<post-id>/review-card.json
raisesawait-signoff
may touchposts/**, rights/** read-only
4Publishing
subagentpublish-run
writespublished/<post-id>.json
raisespublish-failed-twice
may touchpublished/**, posts/** read-only
5Community Engagement
GATE
subagentengage-run
writesconversations/<thread-id>.json
raisessentiment-escalation
may touchconversations/**, answerbank/** read-only
6Month Review
GATE
subagentmonth-close
writesinsights/<month>-report.json, rules/proposed/<change-id>.json
raisesaccept-rule-change
may touchinsights/**, rules/**, published/** read-only, conversations/** read-only

Stack

Every system this agent reads or writes.

System
Read at
Stations
Instagram
Calendar Planning
2 of 6
TikTok)
Calendar Planning
1 of 6
X
Calendar Planning
1 of 6
X and TikTok APIs
Publishing
1 of 6
comments
Community Engagement
1 of 6
each network's own analytics
Month Review
1 of 6
last month's report from station 6
Calendar Planning
1 of 6
mentions and DMs across the four networks
Community Engagement
1 of 6
the LinkedIn
Publishing
1 of 6
the approved answer bank
Community Engagement
1 of 6
the asset library
Post Drafting
1 of 6
the calendar from station 1
Post Drafting
1 of 6
the content calendar
Calendar Planning
1 of 6
the creator agreements
Brand Sign-off
1 of 6
the four networks the brand runs (LinkedIn
Calendar Planning
1 of 6
the monthly report
Month Review
1 of 6
the posted community rules
Community Engagement
1 of 6
the product release notes
Post Drafting
1 of 6
the publishing log from station 4
Month Review
1 of 6
the publishing queue
Publishing
1 of 6
the sign-off queue
Brand Sign-off
1 of 6
the sign-off record from station 3
Publishing
1 of 6
the support desk
Community Engagement
1 of 6
the usage-rights register
Brand Sign-off
1 of 6
the voice guide
Post Drafting
2 of 6
the voice guide and calendar rules the proposals would change
Month Review
1 of 6
Next Step

Want to see if this pattern fits your social automation?

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