Post

Violeta Maftei is a content marketer. While running test prompts for a course, Perplexity named her client's brand in an answer, even though nobody had done any work to put it there. So she built her first automation to see what would happen with effort behind it. 🧵

The client is a small Romanian firm that sells lightning and surge protection equipment. It has been selling for three years, grows on referrals, and does content marketing in bursts. AI search wasn't on the owner's radar at all.

She picked LinkedIn on purpose. Semrush found it's the second-most-cited source across ChatGPT Search, Google AI Mode, and Perplexity, appearing in 11% of AI responses. For professional queries, Profound found LinkedIn is the most cited domain across all six major AI platforms. 🤯

There's even data on what gets a LinkedIn post cited. Scrunch found technical details raise the odds by 77% and named entities by 33%, while Unicode bold formatting lowers them by 58% on ChatGPT. A small firm with deep technical expertise is well positioned for that kind of post.

The stack runs on three tools, each with one job. AirOps orchestrates the workflow and holds the company's Brand Kit and Knowledge Bases. Peec AI tracks which AI prompts mention the brand and which don't. Buffer supplies the LinkedIn engagement data and is where the finished briefs land.

Every weekly run opens by checking what's already in motion. Through Buffer's API, the agent sees what's drafted or queued so it doesn't create duplicate ideas, then pulls reactions, comments, impressions, and reach for published posts. Posts that performed well become signals for follow-ups.

Every gap is then checked against what the company can credibly say. The agent searches 390 older product pages, the firm's social posts, and its YouTube transcripts. Prompts with no supporting evidence get dropped. Only the top five ideas move forward each week. 👀

Those five land in Buffer's Ideas board as writer-ready briefs. Each one carries the AEO prompts the post should answer, the audience, key points, brand positioning, verbatim source material, and the format rules (50 to 299 words, no Unicode bold, direct address).

The writer never touches the automation. They open Buffer, pick up a brief, and write the post. On the next run, the agent checks whether it went live, pulls its metrics, and asks Peec whether the brand is now mentioned for that prompt. That check is what closes the loop.