How the Naming Engine Works

Most startup name generators give you a list and wish you luck. You still have to open 40 browser tabs to find out which ones are available. We do it differently. Here's exactly what happens when you hit Generate.

Step 1: The LLM Generates 30+ Candidates

You give us three things: your idea, your target audience, and a vibe. The moment you submit, a large language model gets to work generating a wide pool of name candidates — usually 30 to 60 of them, sometimes more.

The model isn't just riffing randomly. It's drawing on your inputs to generate names that fit your category, audience, and the feel you're going for. It tries different naming strategies at this stage: invented words, compound words, metaphors, blends, and more. The goal is a diverse pool, not 30 variations of the same idea.

Step 2: The Cliché Filter

Here's where the garbage gets pulled out.

A significant portion of AI-generated startup names sound like they came from the same brain that invented every SaaS product of 2017. Words like Hub, Sync, Nest, Flow, Spark, -ify, -ly — they're generic, they're everywhere, and they make your brand invisible before it even launches.

Our cliché filter removes names that rely on these overused fragments. It also catches names that are structurally identical to existing well-known brands, names that are too generic to trademark, and patterns that research consistently shows audiences find untrustworthy or forgettable.

This step cuts the pool substantially. That's intentional.

Step 3: Phonosemantic Scoring

This one's a bit more interesting. Phonosemantics is the study of how sounds carry meaning and feeling — and it turns out humans respond to the sounds of words even when the words mean nothing.

Hard consonants (K, T, P, hard G) make names sound punchy, fast, confident. Think Kodak, Kraft, Kickass. Soft consonants and open vowels (L, M, N, A, O) make names sound smooth, warm, approachable. Think Alma, Luna, Monzo.

We score every candidate against your vibe. If you chose "Serious & Trusted," names that sound playful or childish get penalized. If you chose "Friendly Weird," we give more latitude. This isn't about being prescriptive — it's about making sure the sound of your name matches the feeling you want to create.

Step 4: Domain Availability via RDAP

This is the part most tools skip — and it's the most important.

Every name that clears the cliché filter and phonosemantic scoring gets checked against the RDAP protocol (Registration Data Access Protocol). RDAP is the system used by actual domain registries — it queries the real, authoritative source for whether a domain is registered or not.

We're not using a cached list, an approximation, or a tool that checks once a week and hopes for the best. We query the real registry in real time, for every name, every generation.

Only names with available domains come back to you.

Why This Matters

Other tools generate names and let you figure out availability yourself. You find a name you love, get excited, then discover the .com has been parked since 2009. That's a bad experience, and it wastes your time.

We only show you names you can actually register right now. The results are smaller than what a tool that ignores availability would give you — but every name on your list is a real option, not a maybe.

The whole pipeline runs in a few seconds. That's what a treat pays for.