Knowledge · Marketing
Am I allowed to, does it work, what does it cost?
The three questions every marketing team asks about AI, in the order they usually ask them. Execution questions deserve numbers, not thought leadership.
Am I allowed to?
The rules, readable
EU AI Act, GDPR, disclosure duties — written so your legal team can read along. That question has its own hub.
Go to /rules/ →
Does it work?
Both directions of the evidence
What shipped and held up — and the documented incidents where it went wrong. We publish both, with sources.
What does it cost?
The question nobody answers
What agentic marketing actually costs to run — models, review time, rework. We are measuring it and publish when we hit ten observations.
The honest version of "does it work".
- Drafting: measured, and strong
- Overnight drafting of campaigns, posts and follow-ups is where teams see the clearest gains — the work exists in the morning instead of being queued behind a person.
- Trust: the actual blocker
- The blocker usually is not the technology. It is that nobody wants to put their name on the output. That is a governance problem, and it has a governance answer.
- Effect claims: mostly folklore
- Percentages travel from deck to deck without a primary source. When we cannot trace a figure, we say so — and we do not print it.
Bring one campaign.
Thirty minutes with someone who built it. We run your campaign through the rules, and you keep the notes.