Week 1: The operating report
I started running my personal brand with AI agents on 6 November.
This is the first operating report, covering activity, distribution, business and quality.
1. Operations
| Channel | Recorded activity |
|---|---|
| 3 posts published, 4 scheduled | |
| LinkedIn articles | 3 scheduled for 11 November |
| X | 5 posts published |
| First post published | |
| Threads | First post published |
| TikTok | First post scheduled |
| Newsletter | 2 editions published |
Time and costs
| Metric | Result |
|---|---|
| My time | About 6 hours |
| Working days | 3 |
| Average time per working day | About 2 hours |
| Claude Max | USD 100 a month |
| ChatGPT Plus | USD 20 a month |
| VPS | USD 6 a month |
| Tool costs in total | USD 126 a month |
The time covers the work recorded through 10 November.
2. Distribution
| Metric | Result |
|---|---|
| X views per post | 6 to 14 |
| X followers | 6 |
| Newsletter subscribers | 249 |
The X account started from zero followers.
These figures are the first distribution snapshot.
3. Business
| Metric | Result |
|---|---|
| Calls booked | 0 |
| Paid jobs | 0 |
A call counts when a calendar slot is confirmed. A paid job counts when its invoice is paid.
4. Quality
How posts are scored
An agent scores each draft against six written criteria.
| Criterion | Points |
|---|---|
| Hook | 0 to 3 |
| Proof | 0 to 3 |
| Pillar fit | 0 to 3 |
| Audience fit | 0 to 3 |
| Sales bridge | 0 to 3 |
| Distinctiveness | 0 to 3 |
| Total | 18 |
Minimum to publish: 11 points.
I review scheduled content and can edit or delete it.
Two agent failures
6 November: A scheduled agent prepared the work but stopped before the final action. Its safety check required human approval.
10 November: A scheduled agent created three LinkedIn articles and their images but failed to schedule them. The task ran in the cloud without access to my signed-in browser.
Both incidents are documented in the failure experiment.
What I learned
The first days produced content across several channels with about six hours of my time.
The two failures showed where the system needs explicit approval and access to the right environment.
The next step is to test the final action, not just the preparation.
Next update
Week 1 ends on 12 November.
I plan to publish the completed Week 1 report on 13 November 2026.
Update history
| Date | Update |
|---|---|
| 6 November 2026 | AI-run brand experiment started |
| 10 November 2026 | Day 5 operating report |
| 13 November 2026 | Next report planned |
Follow the next week
I share the lessons and operating numbers in 5-Minute AI Wins.