Which LLMs does PROEMA optimise for?

Quick answer

Five tracked engines: ChatGPT (OpenAI), Perplexity, Gemini (Google), Claude (Anthropic) and Copilot (Microsoft). Three secondary surfaces tracked: Meta AI, Mistral Le Chat, You.com. Mix is adapted per vertical (B2B, F&B, hospitality) and per language. ChatGPT and Perplexity carry the highest citation weight in 2026.

Which LLMs does PROEMA optimise for, editorially?

ChatGPT (OpenAI) reaches approximately 900 million weekly active users in February 2026 with more than 2 billion daily queries (TechCrunch, 27 February 2026). Two surfaces matter: ChatGPT classic (memory-based, training-data anchored) and ChatGPT Search (live retrieval, uses the Bing index as backend, confirmed by OpenAI late 2024). Both are priority targets.

Claude (Anthropic) is widely used in B2B knowledge work. Two surfaces: Claude classic and Claude with web search. Anthropic uses separate user-agents: ClaudeBot (training) vs Claude-Web / Claude-User (retrieval).

Perplexity is the most aggressive at live retrieval, Perplexity citations are explicit and clickable, making it the easiest LLM to measure citation rate on. It exceeds 800 million monthly queries (Similarweb, March 2026) and is a top priority for premium B2B verticals.

Gemini (Google) powers Google AI Overviews, which surface on the majority of informational queries. The AI Overviews layer is exposed broadly in English and increasingly in French in 2026.

Copilot (Microsoft) is integrated into Bing Search, Microsoft 365, GitHub Copilot. It uses the Bing index as backend, which means a strong Bing presence translates to Copilot visibility. The Bing Webmaster Tools AI Performance Report (launched late 2024) is the only official publisher tool that exposes citation stats.

Apple Intelligence rolls out across the Apple ecosystem since 2024. It partly leans on ChatGPT in fallback mode but is building its own retrieval layer. Worth tracking in 2026.

Secondary surfaces and adaptation logic

Secondary surfaces are tracked by PROEMA in the Continuous protocol: Apple Intelligence (building its own retrieval layer on top of a ChatGPT fallback), Meta AI (rolled out across Facebook, Instagram, WhatsApp), Mistral Le Chat (Europe-anchored, increasingly used in B2B France), and You.com (US-anchored, niche but technical-user-heavy). "The engine mix changes by vertical: ChatGPT, Perplexity and Gemini dominate citations in food and beverage hospitality, Claude joins the top three in B2B technology, and Copilot carries more weight in regulated industries because of enterprise integration," explains Lorenzo Eeman, founder of PROEMA. PROEMA defines the mix at audit kickoff and revisits it quarterly. Sources: TechCrunch (OpenAI WAU February 2026), Similarweb (Perplexity 2026), OpenAI / Anthropic / Microsoft / Google official bot documentation.

At a glance
PriorityLLM2026 anchor
1ChatGPT (OpenAI)~900M WAU, 2B daily queries
2Perplexity800M monthly queries, explicit citations
3Gemini (Google)Powers Google AI Overviews
4Claude (Anthropic)B2B knowledge work
5Copilot (Microsoft)Bing + Microsoft 365 + GitHub
Sec.Apple Intelligence, Meta AI, Mistral, You.comTracked monthly