What does your company actually do?
One answer, then names. Write it the way you would say it to someone who has to understand the business in ten seconds — in nouns, not adjectives.
Concreteness is the whole trick. “An innovative fintech platform” reaches the same abstractions every competitor in your category reaches, and the names come back looking like theirs. “Custody, settlement, ledger” reaches roots nobody else is standing on.
This changes root selection — the concepts matched here decide which Arabic roots are seeded.
generation — changes which names exist
3 more characters. Below three there is nothing to parse — no concept, no root, no register.
Would you like to improve these results?
Fifteen more questions, none of them required, and every one of them states what it changes before you answer it. The names above stay on screen while you work.
Company
What the company makes, and how much work the name has to do on its own. The nouns here reach roots that no adjective can.
What you actually sell or operate. Three is enough — the matcher keeps eight concept slots and the strongest three fill most of them.
This changes root selection. Concrete nouns reach roots that abstractions cannot — this is the highest-yield prose field.
generation — changes which names exist
What the name sits under. Click the selected one again to leave it unanswered.
This changes name length and word count — a sub-brand always spoken after a parent needs less to carry — and stands down the domain factor when no root domain will ever be owned.
generation — changes which names exist
Competitive
Who you will be seen beside, whose shape you admire, and whether you want the ground they are standing on.
One per line. Their names are kept out of concept matching on purpose — feeding them in seeds the territory you listed them in order to leave.
No competitor set. One per line.
This changes how distinctiveness is measured — your competitors become the set every name is measured against, instead of a fixed corpus.
ranking — changes the order of the same names
Brands whose names sound right to you, from any industry. Only the shape is taken.
No name set. One per line.
This changes the target SHAPE — length, syllable count, ending and cluster density. Never the meaning, which would produce imitation.
ranking — changes the order of the same names
Click the selected one again to leave it unanswered, which leaves the territory sort as it is.
This changes whether territories crowded with incumbents are preferred or avoided — the engine already measures occupancy and currently discards it.
generation — changes which names exist
Audience
What the name should leave behind in someone, and where they are standing when they meet it.
24 to choose from. 0 of 5 chosen.
This changes the psychology target every name is scored against.
ranking — changes the order of the same names
Click the selected one again to leave it unanswered, which leaves every factor at its declared weight.
This changes the balance between how a name sounds and how it looks — a name mostly spoken is judged differently from one mostly typed.
ranking — changes the order of the same names
Direction
What the name should be about, how it should be built, and how it should sound. These are the controls with the most reach into what is generated.
Each one names real semantic fields in the root lexicon. Pick as many as you want; each widens the seed set.
This changes which semantic fields of the lexicon are seeded, and those roots count as brief-matched.
generation — changes which names exist
The control with the most reach. Click the selected one again to leave it unanswered.
This changes the pattern set, the register, whether Latin stems are allowed at all, and the target length.
generation — changes which names exist
Measured from the letters of each name, not asserted. Selecting several averages them.
This changes phonetic ranking — measured from the letters, not asserted.
ranking — changes the order of the same names
Letters, not syllables — it is what a client actually counts.
This changes filters the shortlist by letter count, and relaxes itself rather than returning nothing.
ranking — changes the order of the same names
Constraints
What must be excluded, which address you would actually accept, and how much doubt is too much.
One per line. Matched three ways, so a term catches its own near-misses as well as itself.
No word set. One per line.
This changes which names and which ROOTS are excluded — matched on the surface, the folded spelling and the consonant skeleton.
generation — changes which names exist
Which address would actually be signed off. Click the selected one again to leave it unanswered.
This changes which TLDs are checked and how heavily the result counts — an exact-.com requirement ranks differently from “any credible TLD”.
verification — changes what gets checked
Where the screen's threshold sits. The ladder only tightens: there is no rung below permissive, because the grade below fatal is where the religious screen lives and it is never overridable.
This changes the screening threshold. Today every finding graded “serious” survives; the cautious setting promotes it to a rejection.
generation — changes which names exist
Where the mark would have to be registrable. Read the line under this control before you rely on it.
Nothing on this page is a trademark verdict, and no verdict is available from this platform: this is a screen, not a clearance search. A clearance search by a registered trademark attorney covering the relevant Nice classes and territories is required before any name here is adopted, filed or spent against.
This changes which registers we ATTEMPT and how the result is reported. Be clear: every register is key-gated, so the trademark factor still scores zero for everyone until credentials exist.
verification — changes what gets checkedhonest limit — read the sentence above
Why every control here is real
Every brief field is declared in lib/engine/fieldmap.ts with what it changes and the path it takes through the code, and the sentences above are read from that declaration rather than written here. Of 34 declared fields, 29 work today and 1 reaches the engine but cannot be fully honoured — it is the trademark scoping above, and it says so on itself. 4 are retired and are not offered on this form: mission, vision, industry, audience — each of them collected an answer that competed for the same eight concept slots as the description and changed nothing on its own. None collects an answer and discards it.