Ad agent links reporting and creative work

Bernard Aceituno describes handing Google, YouTube and LinkedIn ad performance to a personal AI agent. It tells him which ads are performing, which aren’t, and the pros and cons. He then gives it a video and asks it to build and optimise a new ad, though the post excerpt cuts off before explaining that workflow in full. There are no measured results in the excerpt, so this is an example of a way of working Aceituno says he uses, not evidence of improved ad performance. Read Aceituno’s post.

The useful distinction for an operator is between a one-off task and work that’s complex, repeatable, or uses company data. Aceituno says he uses a personal agent for one-off tasks involving data he considers safe to share, and StackAI for the other work. He lists prospecting, proposals, questionnaires, customer success analytics and forecasting as examples of workflows built there. That’s his stated approach, not a general rule every business should follow.

For a small team, the practical question is where ad review ends and creative production begins, and what information the agent needs to do each job. Start by defining what data is safe to share and what output a person must review. An AI command centre is one option to consider when mapping where agents fit into recurring work.