How GenName.io Works

What a generator actually is

Each of our generator pages is a specialised form and prompt built for one topic — baby names, D&D characters, coffee shop names, and hundreds more. Fill in a few fields (style, origin, tone, whatever applies) and it produces five names at a time, each with a short description and a pronunciation hint. The fields you pick aren't just flavor text — they're turned into explicit constraints the model has to follow, not vague suggestions.

Separately, our tools section (a random name generator, a name wheel, a nickname combiner, and others) runs entirely in your browser. No AI call, no wait, no server round trip — just instant, deterministic utilities.

The AI behind the names

We run open-weight language models, primarily on hardware we operate ourselves rather than renting capacity from a big cloud AI vendor. We don't publish which specific models or hosts we use underneath. When our own hardware is busy or briefly unreachable, a request quietly falls over to a cloud-hosted backup instead of failing, and gets queued rather than dropped if there's a moment of contention.

If you generate names anonymously, we don't tie that output to your identity. If you're signed in, your generated batches are saved to your account's generation history so you can come back to a past search — that's a feature you can see in your dashboard, not hidden logging.

And the AI gets things wrong sometimes. Names can repeat across batches, an AI-suggested name can already be in heavy use, and a model can miss a real naming convention for a culture or language it wasn't trained deeply on. That's exactly why generator pages carry a fixed, sourced list of real names above the essay — checked by a person before it goes live, not generated per pageview. That's the next section.

Real names first, sourced

Since September 2026 we have been rebuilding generator pages to lead with tables of 30 to 80 real names, grouped by sub-type, each with a one-line meaning or context — attested names for a culture or language, canonical names for a game or franchise. Those lists are drafted with AI help against named references, then checked by a person before publishing; they are not produced live by the model. Pages that have been through this pass carry a visible “Updated” note at the end; the rest are being worked through, highest-traffic pages first.

Below the names is a short essay, and any page that makes a factual, historical, or cultural claim ends with a Sources section linking two to four references we actually used — places like Behind the Name, Wiktionary, a country's statistics office, or a franchise's own wiki. Every one of those links is checked before the page goes live, and when a page is materially updated it carries a visible “Updated” note saying what changed.

The full review checklist behind that — what gets checked, by whom, and how corrections get handled — is on the Editorial Standards page.

Where the trend data comes from

The name trends section isn't AI-generated at all — it's built offline from official government birth-registration data, and the baby name and decade generators pull real popularity numbers from the same dataset instead of guessing:

• United States — the Social Security Administration's national baby name data, 1880 to present (ssa.gov)
• England & Wales — the UK Office for National Statistics, 1996 to 2021
• France — INSEE's Fichier des prénoms
• Ireland — the Central Statistics Office's Irish Babies' Names
• Australia — the NSW and Victoria births, deaths & marriages registries, combined

We rebuild this dataset whenever the source publishes a new year, not on any fixed schedule.

Domain, handle, and trademark checks

When you check a name against a domain, we run a real DNS lookup backed by a WHOIS query against that domain's actual registry — not a database of guesses. Handle checks work the same way: we query each social platform directly to see whether that exact handle already exists. The trademark check runs the name against live U.S. federal trademark records through a third-party trademark search API, and flags it as low, medium, or high risk based on exact and similar active matches.

All three are point-in-time signals, not guarantees. A domain can sell in the next hour, a dormant handle can get reactivated, and a trademark search can miss an unregistered “common law” mark. The registrar, the platform, or a trademark attorney is always the final word — not us.

What we do with your data

Generation requests are rate-limited per visitor, which exists to keep the generators available for everyone rather than to track any one person. Saving names into a collection needs a free account through Google sign-in; generating and checking names doesn't require an account at all. We don't sell your personal data — the full detail is in our Privacy Policy.

Where we get things wrong

Every article has a “Report an error” link at the bottom that takes you to our contact page. When a report comes in, we check it against the page's sources — or find new ones if the claim needs them — fix the page, and note the fix at the end of the guide. If something here looks wrong to you, we'd genuinely rather hear about it.

Who runs this

GenName.io is built and maintained by Thien Nguyen. More on why the site exists is on the About page.