All posts
Product 17 December 2025 4 min read

Discovering ENS domains with natural language: introducing AI Search

Find a name by describing what you want, not by guessing strings. AI Search runs semantic similarity across millions of registered .eth names so the right ones surface from a plain English prompt.

Buying or registering an ENS name has always been a guessing game. You think of a word, try it, get told it's taken, think of another, repeat. The names that would actually suit your project — that you would never have guessed — stay hidden in a registry of millions of strings.

AI Search changes that. Describe what you want in plain English. The semantic index returns the closest matches across the full ENS registry — including names that share meaning rather than letters.

How it works

We index every registered .eth name as a vector embedding. When you type a query — "names that sound like a DeFi protocol", "Italian first names", "short food words" — we embed the query the same way and return the nearest neighbours. The result is a list of names that share meaning, not just spelling.

What it is good at

  • Brand discovery. "Modern fintech names ending in -ly" returns dozens of plausible options you would never type into a checker manually.
  • Concept clustering. "Names that feel like a wildlife conservation project" pulls a cohort of nature-themed handles across many lengths.
  • Cross-language nudges. Query in English, get translations and cognates you might not have considered.
  • Availability-aware. The result list is filterable by registration status, length, and digit/letter composition.

Try it

Open /find/search and paste a description of the name you wish existed. The semantic search runs against the live ENS index and most queries return in under a second.

Related — also tagged Product