the research desk for names in the age of ai

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.

research your name ↗
First read on us. Then 50¢ each · no subscription · unlock only what you want.
same name · different frame

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.

if you're building
High recall is a problem.

Models already associate this name with something else. Your brand will compete with their memory every time a user types the query.

if you're monetizing
High recall is an asset.

Traffic type-ins are real. Models emit this domain on their own. The same memory that hurts a new brand feeds an existing one.

if you're investing
Price it on the signal.

Attractor rate plus hallucination calibration tells you what the name is earning today and what it could earn when the current owner moves on.

if you're exploring
Keep the context.

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.

pricing

50¢ per read.

Pay as you go. No subscription. Buy in packs — the larger the pack, the cheaper the read.

starter
$5
one-time
10 full reads · 50¢ each
start with 10 reads
regular
$15
one-time
50 full reads · 30¢ each
start with 50 reads
desk
$40
one-time
200 full reads · 20¢ each
start with 200 reads
what's in every read
Signals you can't get anywhere else.
  • Attractor probenamedesk-only
    We ask the models about a topic, not your name, and measure how often they emit your domain unprompted. Nobody else measures this.
  • Hallucination calibrationnamedesk-only
    We ask each model what it "knows" about the name — and catch when it confidently invents history the name never had.
  • Prompt collision mapnamedesk-only
    Which 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-only
    Claude 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-only
    Same nine signals, four framings — build / monetize / invest / explore. The verdict paragraph is written for the intent you picked.
  • Valuation from live LLM researchnamedesk-only
    The price tier comes from what the models currently know, not a blind average of stale registrar comps. Fresh every read.
  • Brand signals, done properly
    Nine evaluations — recall, confusion, brand, voice, pronounce, typeable, memorability, fit, TM risk — each with its own algorithm and source of evidence.
honest fit

Built for a specific reader.

We'd rather be the right tool for two people than a blunt tool for everyone.

built for
  • 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.

not built for
  • 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.