After writing the story, helping people find it
I asked an AI agent to turn our website diary into different posts for X and LinkedIn, then keep track of what was actually published.
Once we started writing about how we use AI at Chloe’s Mama Kitchen, I also wanted people to find those stories. In September 2026, I asked how we could bring more visitors to the website. The next job was to take an article we had already published and give people a reason to open it.
I asked the agent to help with my own X and LinkedIn accounts, and to make the sharing a regular task. Cindy’s Facebook page stayed with Cindy. Her customers already use it to check the food she is preparing, so our story-sharing workflow had a separate, clearly defined home.
The logo story gave us an early example. On X, the agent wrote a short introduction: Chloe drew the logo, AI helped refine it and recreate it in Figma, and that became the starting point for our design system. On LinkedIn, it gave more room to the steps, Alex’s refinement and Cindy’s direction. Both posts linked to the same full article. The longer post also explained that the cover was an AI story illustration, rather than Chloe’s original drawing.
We made the agent check that the article was really live before sharing it. After posting, it checked the account, the published text and the permanent link, then saved that record. The two logo posts were published on 16 September. By 3 October, the same routine had also shared our magazine story on both accounts.
I also wanted the routine to avoid repeating itself. It checks earlier posts and its saved records, and limits story promotion to one post per account each day. If a publishing result is uncertain, the next step is to look for the actual post before trying again. For a small business, remembering what has already happened is part of doing the job properly.
We added a way to check how people reach the website and whether they open a story or continue to the menu. Those signals need care: reaching the last paragraph does not prove someone read it, and opening the menu does not prove they ordered. Our own checks can also appear in the traffic figures. I keep those limits visible rather than calling every visit a new customer.
What we have completed so far is a repeatable path from a live article to two different social posts, with published links we can check later. We have not established extra sales from it. The useful change is that writing the story, adapting it for each account and keeping its publication record can now sit in one regular workflow.
