AI name generator with a registry check wired into the output
A language model is very good at inventing names and has no idea which of them are already registered. Pairing it with a live registry check is what turns an entertaining list into a usable shortlist.
| Model | Runs server side, your brief is never posted to the browser |
| Constraint | A candidate is only shown once it has been checked |
| Training | Your briefs and shortlists are not used to train anything |
| Fallback | A local generator runs the worked example, so a page view costs nothing |
No card to create the account. The tool below runs without one.
Run it on a brief like yours
The example below is already running against the real registry. Replace it with your own line and the model takes over from the local generator.
Availability is checked live against the registry. Your naming brief stays yours.
| Brief | Worked example, 91 characters |
| Style | Brandable, checking .com |
| Registry | RDAP, one live lookup per row |
| Progress | Starting the example 0 of 5 checked, free to register |
A worked example, running on the brief in the field opposite. Write your own and the names come from the model instead of the fallback generator.
The example brief is being turned into names. Each one is then looked up against the registry that holds the record, and the verdict lands on the row.
On this page
- 01 What the model is good at, and what it cannot know
- 02 How to brief a model so the names are not generic
- 03 What happens to what you type
Every plan is paid and starts at $39 a month. The tool above runs without an account.
What the model is good at, and what it cannot know
Given a decent brief, a model produces name candidates with a range and a fluency that rule-based generators never had. It understands that a physiotherapy scheduling tool should not sound like a nightclub, and it can hold a tone across eight suggestions.
What it cannot do is tell you whether a name is registered. It has no live view of any registry, and when asked it will produce a confident answer shaped like the truth. Every AI naming tool that skips the check inherits this problem and passes it to you.
So the model writes and the registry decides. The generation step produces candidates, the RDAP step produces verdicts, and only rows with a verdict are shown as results.
How to brief a model so the names are not generic
Write a sentence, not a keyword. Keywords produce the average of everything ever written about that category, which is precisely the pile of names you have already rejected.
Include the constraint that makes the business unusual. Not the industry, the difference: by post, for older houses, unscented, seasonal crews, insurance codes. Those details are what pull the output away from the average.
Say how it should sound if you know. Plain, clinical, warm, blunt. Tone in equals tone out, and it is much faster than rejecting forty names in the hope that the tool guesses.
What happens to what you type
The brief goes to our server and from there to the model. It does not go to the browser of anybody else, it is not sold, and it is not used as training data for us or for anyone we work with.
The worked example that runs when the page loads deliberately skips the model and uses a local generator, because a page view should not spend a generation. The registry checks on that example are real; only the invention step is local.
The same engine is framed two other ways on this site. The brand name generator leans on distinctiveness for a company that intends to defend its name, and the domain name generator starts from the registry answer and works backwards. One brief works on all three.
Questions people ask first
The rest are on the FAQ, and the mechanics are in how it works.
Not on its own. A model has no live connection to any registry, so an availability answer from a chat interface is a guess dressed as a fact. Here the model only invents; the availability answer comes from an RDAP query against the registry.
A hosted general purpose model, called server side so the key is never exposed to the browser. The specific model matters less than the constraint around it: no candidate reaches your screen without a real check attached.
No. Briefs and shortlists are yours. We do not resell them and we do not use them to train anything, which also means we cannot recycle one client's brief into another client's shortlist.
Generation is not deterministic, and that is useful: a second run on the same brief is a second opinion rather than a repeat. Names you want to keep should go on a shortlist rather than be relied on to reappear.
Close to this one
Three pages that answer the neighbouring question, each with the console seeded for its own kind of business.
Most domain name generators are word games with no connection to a registry. This one treats availability as the constraint rather than an afterthought, so a suggestion only becomes a suggestion once the registry has answered.
A brand name has to be ownable, and ownable means two things at once: distinctive enough to protect, and available enough to register. This generator refuses to show you names that fail the second test.
App names live in a store listing, on a home screen and in a search box, all of which punish length. This generator leans short and checks every candidate against the registry as it goes.
Write one line and see what survives the check.
Every name that comes back free has been looked up against the registry that holds the record. Nothing on the shortlist is already gone.
No card to create the account. Your brief stays yours.