Agent to track whether AI recommends a client or just mentions them
Hi everyone Built an agent at the cohort in London last week. The idea came from 2 or 3 calls where this was requested as a KPI. The agent checks whether AI actually recommends a client or just mentions them [or not that they're absent] linkedin.com/feed/update/urn:li:activity:7522012950329479168 For every AI answer, it ranks the brands in the order AI lists them and scores where the client lands. It outputs a sheet with one row per answer:
Prompt: the question a buyer asked AI, from the client’s tracked prompts.
Engine: which AI answered (ChatGPT, Perplexity, Google AI Overviews and so on).
Brand ranking: the client’s spot in the answer (#1, #2, #3…) against the other brands AI named, in the order AI listed them.
Status: one of the four scores: First choice, Recommended alongside, Mentioned not recommended, or Absent.
Evidence: the exact sentence AI wrote about the client, checked word for word.
Why the competitor won: the reason AI preferred the other brand, usually price, features, ease of use or team fit.
Cited sources: the sites and pages AI pulled from for that answer.
Attack flag: Attack if it’s a gap content can realistically close, Do not attack if the client already wins or the competitor wins on a fit the client doesn’t go after.
Added this to 3 of our clients so far. Very positive reaction. Great value add and useful to inform content. Attached is the template if you want to use it