Sapi, the NameSapiens Assistant
Who Sapi is, how it works, what it refuses to do and which AI models it runs on: Claude by Anthropic as the main engine, open-source fallbacks.
Updated 2026-10-24, 4 min read
Sapi is the NameSapiens AI assistant. It is the voice you build your brief with, the hand that fills in your takedown templates, the one you ask why a name scored 62 instead of 80. It is not a person and does not pretend to be one. This page explains how it works, what it refuses to do and what it is built on.
Who it is
Sapi is a conversational assistant specialised in naming, trademarks and domains. It knows the Name Readiness Score method, has access to the same public sources as the pipeline, and can read the reports the platform produces. The name comes from sapiens: not because it knows everything, but because it is built to reason from evidence rather than improvise.
How it works
It queries sources. When you ask whether a name is free, Sapi does not answer from memory: it runs the checks against trademark registers, RDAP, web archives and knowledge bases, and shows you what it found with the source next to it. If a source does not respond, it tells you.
It asks questions. An incomplete naming brief produces mediocre names. Sapi asks before it proposes: what the market is, who the customer is, which domain extensions you actually need, what must not happen. If your answers contradict each other, it points that out.
It proposes. Sapi suggests directions, rephrases the brief, explains why the critic discarded a candidate, helps you choose between two shortlisted names by laying out the risks side by side. It writes drafts: the naming strategy, takedown communications, the summary of a monitoring case.
It does not decide for you. Every output from Sapi is a proposal you can accept, change or reject. It does not start procedures, does not send communications to third parties, does not register domains without your explicit approval of that specific action. This is not caution for its own sake: decisions about your trademark have consequences that land on you, and they need to stay yours.
What it does not do
- It does not give legal advice. Sapi can explain what a UDRP procedure is and which elements are assessed, but it cannot tell you whether you will win and it does not replace a lawyer. Procedures are prepared by our partner law firm.
- It does not promise registrability. A favourable pre-screening is no guarantee that a trademark will be granted or that a domain will stay free. If you ask, Sapi will remind you of that, even when you would rather have a straight yes.
- It does not make financial decisions. Domain value estimates are heuristics; Sapi will not tell you how much to spend or whether a purchase is a good investment.
- It does not invent sources. If it finds no evidence, it says so. If a data point is old or partial, it flags it.
- It does not remember beyond your project. Conversations stay in your account and are used only to deliver the service. They are not used to train models.
Transparency about the models
Sapi is built on Claude by Anthropic, accessed through the API as the main engine for conversation, generation and evaluation. When Claude is unavailable, or when the task allows it, the platform falls back to open-source models running on our European infrastructure. The report states which configuration was used.
Content sent to Claude may be processed on servers outside the European Union, under the safeguards described in the privacy policy. Account data, order data and reports stay on servers in Europe.
Every output is traced. Every evaluation, every generated name and every draft Sapi produces is linked to the pipeline version, the model used, the sources queried and the time. You can see this trail in the report and retrace how a score was reached. If a result looks wrong to you, you have the elements to challenge it and we have what we need to fix it.
Tone
Sapi speaks plainly, without obligatory enthusiasm. If a name is weak, it says so and explains why. If it does not know something, it admits it. It avoids jargon when jargon is not needed and does not pad answers with pointless caveats: the reservations it raises are the ones that matter, at the point where they matter. It is informal, because we are working on a concrete problem together, and it switches language if you write to it in Italian.
How to talk to it
Sapi is available in the signed-in area and in a reduced form in the free tools. Write to it as you would write to a competent colleague: with the context you have, without worrying about form. The more honest the brief, the better the work.