This week in numbers (real, from our system)
- 🤖 AI agents running: 19
- 📝 Content published: 82 (blog RU 20, EN 30, Altezza 32)
- ⚙️ Generated programmatically: 4
- 📥 Leads in the system: 363 (+1 in the last 7 days)
Short answer: Our AI agents draft articles, qualify leads, build SEO pages and read analytics, while people set direction and approve what ships. Building in public means we show how this content automation really works, including what breaks. The figures are in a separate block.
Most posts about AI agents show the polished demo. I'd rather show the workshop. This is the weekly diary of how our content factory runs, written by the people who maintain it.
What does building in public mean for us?
It means no highlight reel. Each week we write down what the agents did, what went wrong and what we changed. We don't put numbers in the text. The system adds them as a separate block, so the prose can't drift into hype.
I do this for a practical reason. When you have to explain a failure in writing, you fix it properly.
How do AI agents run the content factory?
Think of a small production line. Each agent has a narrow job and a clear handoff.
- Writing: one agent drafts articles from a brief and the keyword plan.
- Sales: another qualifies inbound requests and prepares replies for a human to review.
- SEO pages: a generator builds structured pages from verified data.
- Analytics: an agent reads the results and flags what to change next week.
What breaks every week?
Look, things break. Drafts drift off-topic. Agents get too confident about facts they can't verify. A small change to one prompt can quietly change the output of the whole line.
Our fixes are dull and they work. We add validation rules, ban fabricated data, and make every agent say "no data" instead of guessing. That last rule saves us the most pain.
Where do people stay in the loop?
Everywhere that judgment matters. People choose topics, set the tone, approve what gets published and talk to clients. The agents take the routine: first drafts, formatting, repetitive checks. That frees us for the work only humans can do. We build this to augment the team, not replace it.
Frequently Asked Questions
Do AI agents replace the marketing team?
No. They remove repetitive work. Strategy, positioning and final approval stay with people.
Why share the process publicly?
Because honest detail builds more trust than claims. It also forces us to keep the system clean.
Can a small business build a similar content factory?
Yes, if it starts small. Automate one routine task, add human review, and expand only after it works reliably.
Why are there no numbers in this diary?
We keep metrics in a separate, system-generated block. That way the text stays honest and the data stays verifiable.
Conclusion
A content factory run by AI agents is mostly plumbing, review and patience. We'll keep documenting it week by week, mistakes included. If you want to build something similar for your business, get in touch and we'll talk through where to start.