Peec AI

Is there a recipe for getting your brand cited in AI search, or is everyone just guessing? Last week we hosted a fireside chat at Berlin's Flussbad Campus to tackle the questions splitting the field right now. The panel dug into the state of AI search, debunking myths around Reddit, and what tactics are currently paying off.  Here are five takeaways shared during the discussion, keep an eye out for the full length version coming soon to YouTube. 1️⃣ One panelist said Reddit spam is largely a waste of time because it is difficult to influence and rarely yields immediate ROI. 2️⃣ Track brand perception rather than just presence, as entrenched AI knowledge shifts slowly. 3️⃣ Focus on solid SEO, which covers half of your needs, while using AEO to provide additional value. 4️⃣ Avoid tracking prompts like keywords, and instead select specific topics based on what customers type into ChatGPT. 5️⃣ One panelist noted that Facebook and YouTube videos yield significantly more citations than TikTok experiments. Taking place in the Reethaus hall, the discussion featured Ethan Smith, CEO of Graphite; Niklas Buschner, founder of Radyant and host of the Masters of Search podcast; and Kira Hankamp (geb. Chaluppa), Head of SEO at idealo. They were joined by Malte Landwehr, Peec AI's CPO and CMO, and moderated by Antonia Breitenfellner, AI Search Strategist at Peec AI. Thank you to the panelists, everyone who attended, and the team at Flussbad Campus. The full discussion lands on YouTube later this month, follow for more updates www.youtube.com/@Peec-AI

My new research is out. It is about a small quirk I found in our research corner at Peec AI. The words in your company name influence how AI assistants describe you. 904 real brands. 5 AI assistants. 10 attributes each. I scored every attribute against a dataset of 14,000 words that humans rated from 1 for negative to 9 for positive. The effect showed up in live data with web search on, and in a controlled study with it off. On average the impact is small. The more interesting finding is that the model sometimes describes the word and not the company. 1️⃣ A fictional company called "Unethical Inc." was described as "Dishonest, Corrupt, Deceptive." "Merciless Inc." got "Ruthless, Cutthroat, Relentless." 2️⃣ On GPT 5.6 Terra, that happened in one description out of twelve. 3️⃣ Luckily, famous brands are immune. No AI assistant tells you that Slack ("slack" 3.85/9) is loose and lazy. The effect only shows up on brands the AI models don't recognize. So, when you name a new company, learn from Lovable ("lovable" 8.26/9). And if it is too late for that, get known, so the assistants stop confusing your company name with the word itself. Why you could be a dolphin instead of a donkey, and much more, is in the full blog post. Links in the comments. All the thanks for the help to the growing GEO team: Tomek Rudzki, Malte Landwehr, Metehan Yeşilyurt and David Konitzny.

New research just dropped @Peec AI Rerankers don't get talked about much in the GEO/AEO world yet. I think they deserve more attention, because they are one of the models involved in deciding whether AI search cites you. Quick intro if the term is new: when an LLM answers a question, it first retrieves dozens or hundreds of candidate passages from the web. Then a reranker, a smaller model specialized in relevance, reads each passage against the query and scores it. The strongest candidates move on to the LLM writing the answer. It sits between "retrieved" and "cited." I was curious how this stage actually behaves, so I ran 12 open-source rerankers on the same two passages. Passage A was a solid product description. On-topic, well-written, relevant. Passage B was one sentence: a direct shortlist answering the query. The scores surprised me: 📍 MiniLM gave Passage A 0.0019%. Passage B: over 99.9%. 📍 BGE-large: 0.20% vs 99.8%. 📍 Qwen3-Reranker: 0.27% vs 99.9%. Same topic. Same brand. Very different outcome. My takeaway: "about the topic" and "answers this query" are two different tests. A lot of good content passes the first and quietly fails the second. It also 'helps' explain why listicles show up in citations so often. Not because AI loves lists, but because they contain extractable answers, named entities, and coverage for multiple fanout queries. The useful principle is intent-to-answer-shape alignment, not turning everything into a list. And the simplest edit I found: one direct-answer sentence in the first two lines of the section that targets each query. Not a rewrite. One sentence. I wrote up the full guide on the Peec AI blog: how BGE, BERT-family rerankers work, what SPLADE and ColBERT can tell you about your content, what ChatGPT is probably running (and not), and a "which model when" decision guide for GEO/AEO teams. Not only for writing, but also for deciding which models to use for which research. With animated cards showing each model reading the passages in real time. Link in the comments 👇

It's not all GEO research and shipping new features here at Peec AI HQ 👀 It's also summer evenings on the rooftop with some seriously amazing people 🖤

Next up in our agency spotlight – ROAST ⚡ Independent London agency driving growth since 2015, now running full-service GEO for B2B and B2C brands. With Peec AI, they track visibility and sentiment across AI search for clients like AJ Bell, PitchBook, the TUC, and AllSaints, particularly strong in regulated industries like finance, insurance, and healthcare. Need expert help with AI search? Browse our trusted partners directory → peec.ai/agency-directory