You can publish an article a day and ChatGPT will never cite you, while a competitor of yours gets named everywhere despite publishing half as much. The reason is that the AI doesn’t count your content — it counts who talks about you: an industry expert who cites you is worth more than six months of posts on your blog. If no recognized name ever mentions you, to the AI you are simply invisible. Building the right alliances in your field is the move that changes everything — and it doesn’t require budget, only method.
Have you ever noticed that certain names show up in AI answers with a regularity that can’t be explained by the volume of content published alone? Professionals who don’t have the biggest site, don’t publish more, and yet get cited as a reference by the model every time a certain topic comes up.
The explanation might be simpler than you think. And it has to do with who talks about you, not with how much you talk.
When a recognized professional in your field mentions you — in an article, in a talk, in a technical publication — that passage isn’t just a compliment between colleagues. To a language model it’s a structural data point that changes the weight of your name in the vector space. And it does so in a different, and more powerful, way than a generic mention.
The mechanism: how AI distinguishes who mentions you
To understand why a mention from a peer carries more weight, you have to look at how models process the context of sources. It’s not a trivial process: the system evaluates a series of signals that go beyond the textual content.
Srba et al., in their study on source credibility, document this precisely:
“Context-based signals considering user/source cues like domain reputation and publication metadata contribute most towards human judgement.”
The key point is “contribute most”. It’s not the text itself that weighs the most, but the context signals: the reputation of the domain, the publication metadata, and — this is the part that matters to us — who the source speaking is. If whoever mentions you is an entity the model already recognizes as authoritative on a certain topic, the signal it transmits to your name inherits part of that weight.
In practice it works like a network of professional references, but computed on a statistical scale. The model has processed millions of texts during training. If a person who appears frequently in authoritative content on a certain topic mentions you in that same context, your name benefits from that association. Not because the model “decides” that you’re good — but because the co-occurrence with a high-weight entity pulls your vector into the same region of the semantic space.
Why a peer weighs more than a generic media outlet
Here comes a distinction that makes the operational difference. A mention in a general-interest newspaper is useful — I talked about it in the article on implicit reference. But a mention from a recognized expert in your specific field has one extra characteristic: deep thematic coherence.
The reason lies in how embeddings work. Gao et al., in their 2024 work, explain a technical aspect that illuminates this point:
“Sparse and dense embedding approaches capture different relevance features and can benefit from each other by leveraging complementary relevance information.” Gao et al., 2024
If I were to translate this sentence into something directly useful: the system captures relevance on multiple levels. There’s a lexical level — the exact words that appear — and a deep semantic level, the overall meaning of the context. When a peer in your industry mentions you, both levels work in your favor. The industry keywords are the same. The semantic context is the same. The model doesn’t have to make any effort to connect your name to that domain of expertise, because the connection is already inside the structure of the text.
Compare that with a mention from a general-interest journalist who writes “the marketing expert Roberto Serra”. The signal is there, but it’s diluted: the article’s context might be about ten different topics, and your name is one of many mentioned. When instead a colleague who works in your field cites you in the context of a specific technical discussion, the signal is concentrated. Every term around your name reinforces the same association.
But if no peer in your field ever mentions you, you’re building an island.
How this signal propagates in RAG systems
The weight of a peer endorsement doesn’t stay confined to the static training data. RAG systems — the ones that search external sources before generating the answer — amplify this signal significantly.
Preslav Nakov et al., in 2026, describe how this amplification works:
“By maintaining the agentic reasoning loop across KG and web retrievals, our framework enables dynamic, multi-source evidence synthesis.” Nakov et al., 2026
“Multi-source evidence synthesis” is the key. The system doesn’t take a single source and present it as the truth. It cross-references multiple sources, aligns them with the knowledge graph, and builds a synthesis. In this process, if your name appears mentioned by multiple recognized peers — each already present in the knowledge graph as an authoritative entity — the system finds cross-confirmations. And cross-confirmations are exactly the kind of signal that pushes a name toward the final answer.
From this follows a deduction you won’t find in any paper, but that the mechanics make logical: mentions between peers create a network effect. If expert A mentions you, and expert B mentions you, and A and B also mention each other, the model is looking at a dense and coherent subgraph. And the density of a subgraph, in the context of multi-source retrieval, is a reliability signal.
It’s the same principle I analyzed when talking about co-citation: when multiple sources converge on the same reference, the weight of that reference rises. Peer endorsement is the most powerful version of this pattern, because the converging sources aren’t generic — they’re already authoritative on the specific topic.
Write contributions for publications where the other experts publish.
The mistake of building in isolation
Many professionals invest everything in their own content. They publish articles, maintain the site, optimize the pages. And all of this is necessary: topical authority is also built from within. But if no peer in your field ever mentions you, you’re building an island.
Your node in the knowledge graph exists, but it has few connections to the other authoritative nodes. And a poorly connected node, even if rich in content, gets retrieved less often by the system when it has to synthesize an answer. The model prefers nodes that have external confirmations, especially if those confirmations come from sources it has already classified as relevant.
The difference between who gets cited by the AI and who doesn’t often isn’t in the quality of the content. It’s in the network of references. And that network is built with actions that most companies consider “public relations” — reciprocal interviews, content collaborations, panel participations, contributions to industry publications. All activities that generate peer-to-peer mentions.
How to start building the signal
You can do a first check right now. Search for your name on Google in quotes, excluding your own domain. Among the results, count how many come from content written or curated by professionals in your field — not from generic outlets, not from press releases, not from your site. Those are the peer endorsements the model has already processed.
If you find few, you’ve identified the bottleneck. And the good news is that it doesn’t require an advertising budget — it requires professional relationships. Participating as a speaker where your peers participate. Collaborating on industry research. Writing contributions for publications where the other experts publish. Every time one of them mentions you in that context, you’re adding a high-weight signal to the representation the AI has of you.
If you want to understand how your Knowledge Panel comes into play — because being recognized as an entity in the knowledge graph amplifies every peer endorsement you receive — I cover it in the dedicated article. And if you want to dig into how the attention mechanism concretely computes the weight of these associations, it’s a technical step worth taking.
The strongest signal you can send to the AI isn’t talking about yourself. It’s getting the right people to talk about you, in the right context. That’s the mechanics. The rest is a consequence.
