Elastic Owns 83% of AI Search Discovery. Perplexity Appears in 1 in 4 Responses.

When a buyer asks ChatGPT, Gemini, Claude, or Perplexity to recommend "AI search tools," which brands show up?

Not the ones you'd guess.

Elastic — a search infrastructure company best known for the ELK stack and Elasticsearch — surfaces in 83% of AI-generated responses to buyer discovery prompts in the AI search category. Algolia, an API-first search platform, appears in 63%. Glean, the enterprise work AI, makes 49%. Microsoft Copilot shows up in 34%.

Perplexity, the consumer AI search product that has received more press coverage than any other tool in the space, appears in 26% of responses — tied with Weaviate and sitting behind Lucidworks Fusion, Bloomreach, and Constructor.

This is the AI search category paradox: the brand that consumers associate with AI search is not the brand that B2B buyers — or the AI engines answering their questions — associate with it.

The category is not what it sounds like

The word "AI search" has two completely different meanings depending on who is using it.

To consumers, AI search means Perplexity — the product. The chatbot that answers questions with citations. The tool that is replacing Google for a generation of researchers and knowledge workers.

To B2B buyers, AI search means search infrastructure — the engine that powers intelligent retrieval inside a product, a knowledge base, or an enterprise. They are asking "how do I build search that understands intent?" not "where do I go to search the web?"

AI models have learned to distinguish these contexts. When a buyer asks about "AI search tools," the models retrieve results in an enterprise frame: Elastic for infrastructure, Algolia for product search, Glean for workplace retrieval. Consumer AI search products like Perplexity appear in responses to a narrower set of queries about conversational search or research tools — not the broader "AI search" category as enterprise buyers define it.

The result: the brands winning AI discovery in the AI search category are almost entirely infrastructure and enterprise vendors, despite consumer AI search getting the majority of press attention.

Why Elastic leads by such a wide margin

Elastic's 83% share-of-voice is not an accident. Three factors explain it.

Brand-to-query alignment. When an enterprise buyer asks about AI search, they are describing a technical capability — vector search, hybrid retrieval, semantic ranking — that Elastic has built, documented, and marketed for years. Elastic publishes deep technical content on exactly these topics: blog posts on kNN search, tutorials on Elasticsearch's ELSER model, documentation on hybrid BM25 + vector queries. That content is public, indexed, and citable. AI models retrieve it because it is the most detailed, on-topic material available.

Temporal advantage. Elastic has been in search infrastructure for over a decade. Its documentation, case studies, and community content have compounded into an enormous body of AI-citable material. A newer entrant publishing equivalent content today starts from zero. Elastic starts from tens of thousands of indexed, authoritative pages.

Category vocabulary ownership. Ask any AI model what "semantic search" or "vector search" means, and the answer will reference Elasticsearch at some point. Elastic owns the technical vocabulary of search at a level that means any enterprise query touching those terms is likely to surface it. It is the same dynamic that makes HubSpot unavoidable in CRM queries — vocabulary ownership is the upstream driver of citation dominance.

How the category breaks down by topic

The aggregate share-of-voice numbers hide meaningful variation at the topic level. Six distinct buyer intent clusters drive AI search discovery, and brand performance shifts significantly across them.

Semantic and vector search is Elastic and Weaviate territory. Both surface in 4 of 5 responses in this cluster. Weaviate's 26% overall share understates its relevance — it concentrates its citations here, where its vector-native positioning aligns tightly with the query intent.

Workplace and enterprise knowledge retrieval is Glean's strongest cluster. Glean surfaces in 5 of 5 responses in this category, ahead of Elastic. Microsoft Copilot competes here too. This is the one cluster where Glean consistently outpaces Elastic — the use case is Glean's home territory.

E-commerce and product discovery belongs to Algolia, Constructor, and Bloomreach. All three surface in 5 of 5 responses in this cluster. Outside of this topic, their share-of-voice is significantly lower. Their overall percentages reflect strong performance in one specific bucket, not broad category coverage.

Developer search APIs is another Algolia stronghold, with Elastic close behind. This is why Algolia's 63% overall share understates its strategic position: it dominates two high-intent clusters that attract serious buyer evaluation.

RAG and LLM-powered search is the emerging battleground. Glean and Microsoft Copilot surface most often here, followed by Elastic. This cluster is growing as enterprise buyers start asking how to build retrieval-augmented pipelines — exactly the query set that will drive the next wave of AI search evaluation.

Hybrid search and relevance tuning is Elastic's most dominant single cluster. Five of five responses. No close competitor. This is the cluster that anchors Elastic's 83% overall number.

The topic breakdown reveals a pattern that the aggregate numbers obscure: several brands with modest overall share-of-voice own specific clusters completely. A brand appearing in 29% of all prompts but 100% of the prompts in one high-intent cluster is better positioned than a brand appearing in 40% of all prompts with no cluster dominance.

Algolia's 63%: the API-first alternative

Algolia sits in a strong second position. It surfaces consistently across product search, e-commerce search, and developer-tooling queries — all subsets of the AI search category. Its developer documentation is among the best in any software category: structured, use-case-specific, and deeply indexed. AI models retrieve Algolia when they need to recommend a managed search service rather than self-hosted infrastructure.

The gap between Algolia (63%) and Elastic (83%) reflects positioning breadth, not content quality. Elastic covers more query types — infrastructure, enterprise, vector, analytics — while Algolia's strongest signal concentrates in product search and developer tools. Both have strong content. Elastic's footprint is simply wider.

The Perplexity paradox

Perplexity appearing in 26% of AI search discovery responses is counterintuitive until you examine the prompt set.

The 35 discovery prompts in the AI search category are buyer prompts: "what search tools should I evaluate for my product?", "what is the best AI search platform for enterprise?", "how do I add intelligent search to my knowledge base?" These are enterprise procurement queries. Perplexity is a consumer research tool. AI models do not recommend Perplexity when a buyer asks how to add intelligent search to their CRM.

When the query explicitly invokes "AI search engines" or "AI-powered research tools," Perplexity surfaces reliably. But the broader "AI search" category prompt pool is dominated by enterprise and infrastructure intent — which is where Perplexity's consumer positioning creates a natural ceiling.

This is a category-definition problem. Perplexity is the dominant brand in AI-powered web search. It is not the dominant brand in AI-powered enterprise search. Those are different markets, and AI models have learned to treat them accordingly.

What this means if you compete in this category

If you are Elastic or Algolia: your AI visibility is strong, but it is not permanent. The gap between Elastic (83%) and the brands at 25–34% is wide enough that a sustained content push from a competitor could close it. Stack your defensible signals — case studies with named customers, third-party integrations, analyst recognition — to make your position harder to displace before a challenger does.

If you are Glean, Bloomreach, or Constructor: your 29–49% share-of-voice means you are consistently in the conversation but not consistently first. The topic breakdown matters more than the aggregate. A brand at 29% that dominates one high-intent cluster — as Constructor and Bloomreach do in e-commerce — is better positioned than a brand at 49% spread thinly across every topic.

If you are Perplexity: the AI search category as defined by enterprise buyer prompts is not your market. Competing for citations in a category dominated by infrastructure brands requires either building a credible enterprise content presence or defining a distinct sub-category where your consumer positioning becomes an asset rather than a liability.

If you are a challenger brand not appearing in the top 10: this category has meaningful mid-tier activity. Appearing in 3–5 prompts out of 35 does not register in share-of-voice, but it identifies which prompt types your content is beginning to reach. That is the signal to build from.

The counterintuitive lesson

The brands winning AI search discovery built their positions before "AI search" was a category. Elastic's citation dominance is the accumulation of a decade of search infrastructure content. Algolia's 63% is the result of developer documentation that has been compounding for years.

This is the core challenge for newer entrants and consumer AI products trying to enter the enterprise frame: you are competing against companies that started building their AI-citable content footprint in 2014. The content clock does not reset when a new category name emerges.

The AI search category is being defined right now by the models answering buyer questions. The brands publishing technical content, releasing case studies, and owning the vocabulary of search are the ones the models cite. In May 2026, that is Elastic and Algolia. The question for every other brand in the category is what their content footprint looks like twelve months from now.

---

Data from the unCited.ai AI Search category — 35 discovery prompts and 80 head-to-head evaluations across ChatGPT, Gemini, Claude, and Perplexity. 1,000 total citations captured. Data refreshed May 7, 2026. Explore the full category →