Most marketers still optimise for “AI visibility” as if large language models (LLMs) share a single index.

They do not.

Each assistant has its own ingestion pipeline, retrieval weighting, and update cadence. Treating them as one channel is equivalent to running the same media plan across Google, Meta, and TikTok without accounting for format or audience.

The result: wasted spend, inconsistent presence, and unmeasurable leakage.

This seem relevant for users as well.

Oh no. They've already started to figure out how to advertise to AI. I predict reduced AI accuracy in our future.