Optimizing "for AI" in the abstract does not work, because each assistant builds its answer differently.
ChatGPT
Blends trained knowledge with browsing. On generic queries it leans on what it learned: the brand with the strongest historical web presence wins. When it browses, it cites few but highly influential sources.
Implication: work on long-term brand recognition and get onto your category's reference pages.
Perplexity
The most literal one: it almost always retrieves live and shows numbered sources. The answer is essentially a summary of the 5-8 URLs it picked.
Implication: the most "hackable" assistant with well-structured, recent content. Also the best thermometer for which sources dominate your category.
Gemini
Leans heavily on the Google ecosystem: indexed content, business profile, reviews and structured data. Usually the most sensitive to local signals.
Implication: if your sales have a geographic component, business profile and reviews remain decisive.
What to do with these differences
- Measure each assistant separately. A global average hides that you are strong in one and absent in another.
- Use Perplexity as a lab: content changes show up there in days, not months.
- Use ChatGPT as a brand thermometer: if you appear without browsing, your structural presence is solid.
- Use Gemini to audit your Google signals.
A mature AEO strategy is not about "showing up in AI": it is about closing the specific gap in the specific assistant where you are losing.