Domain researchfor the moment the LLMsare doing the searching.
Your name isn't competing with other names anymore — it's competing with the model's memory. Namedesk tells you how a domain actually performs when Claude, GPT, and Gemini are the ones deciding what your user sees.
See what a full read looks like.
Every sample is a real report — the same 9 signals, the same verdict paragraph, the same pricing band you'd get running your own name. Click into any of them.
Saturation-level recall. Every model knows it cold — see what the report does when the name is the category.
Clean brand, high attractor rate. The shape of a name the models reach for first when the topic comes up.
A dictionary word on an alt-TLD. Watch how the models confuse the brand with the concept.
An adjective on a product-TLD. See the intent-routed verdicts side by side.
Same name. Four verdicts.
The signals don't change. What they mean does. A name that's right for a founder is often wrong for an investor, and vice-versa. Every report routes through your intent.
Models already associate this name with something else. Your brand will compete with their memory every time a user types the query.
Traffic type-ins are real. Models emit this domain on their own. The same memory that hurts a new brand feeds an existing one.
Attractor rate plus hallucination calibration tells you what the name is earning today and what it could earn when the current owner moves on.
Adjacent names often score better for the same intent. We flag the near-misses and prompt-collisions so you can pivot to a cleaner variant.
50¢ per read.
Pay as you go. No subscription. Buy in packs — the larger the pack, the cheaper the read.
- Attractor probenamedesk-onlyWe ask the models about a topic, not your name, and measure how often they emit your domain unprompted. Nobody else measures this.
- Hallucination calibrationnamedesk-onlyWe ask each model what it "knows" about the name — and catch when it confidently invents history the name never had.
- Prompt collision mapnamedesk-onlyWhich queries drag your name into a different brand's answer? Which neighbours are you competing with in the model's memory?
- Recall across four frontier modelsnamedesk-onlyClaude Opus, GPT-5, Gemini 3 Pro, and Kimi K2 — in parallel, on every read. One model's answer is an anecdote; four is a signal.
- Intent-routed verdictnamedesk-onlySame nine signals, four framings — build / monetize / invest / explore. The verdict paragraph is written for the intent you picked.
- Valuation from live LLM researchnamedesk-onlyThe price tier comes from what the models currently know, not a blind average of stale registrar comps. Fresh every read.
- Brand signals, done properlyNine evaluations — recall, confusion, brand, voice, pronounce, typeable, memorability, fit, TM risk — each with its own algorithm and source of evidence.
Built for a specific reader.
We'd rather be the right tool for two people than a blunt tool for everyone.
- Founders naming a company
You're about to commit to a brand for the next decade. We tell you which names the models will confuse with yours, which ones they've already associated with something else, and which ones are clear of that memory.
- Investors valuing a name
A valuation built on live LLM research — attractor rate, hallucination profile, recall distribution across four frontier models — not a blind average of stale registrar comps. The signal the market hasn't priced yet.
- Bulk flippers and registrar farms
We don't do batch uploads, API resellers, or pay-per-thousand lookups. This is a research desk, not a pipe.
- SEO-first namers
If you're buying an exact-match keyword domain for search juice, there are cheaper tools. Our signal is brand fit in the LLM era, not SERP position.
What do the models already think of your name?
First read on us. 50¢ each after that. No subscription, no catch.