How Word Says Hi works

Word Says Hi returns a generated, provisional word history for a submitted word. This page explains how a lookup travels, what the labels mean, and how the curated pages are made.

How a reading is produced

You type a word (up to 80 characters) and an optional language hint (up to 100 characters). Nothing is sent while you type. Selecting Trace this word sends the pair to this site's Cloudflare Worker, which checks the request, applies an abuse limit (10 lookups per 60 seconds per connecting IP at each Cloudflare location) and calls Qomza's analysis gateway through its /v1/analyze contract. The gateway asks a model for a structured answer that must match a strict JSON schema and repeat your exact word; the Worker rejects responses that do not. The reading is then drawn on the page as a route, meaning shifts, related forms and uncertainty notes.

Each lookup also carries technical request metadata — the IP address Cloudflare sees for your connection, an approximate browser language preference, a per-request identifier and app details — described in full on the Privacy page.

Readings are generated by a model

Every reading is written by an AI model, not assembled from cited sources. It can be wrong, incomplete or misleading, and it supplies no verified citations. Treat it as a starting point for your own checking, not a verdict — the Terms and the reading itself say the same.

How uncertainty is shown

Uncertainty appears in four places. The status line reads “A path is suggested”, “Some steps are uncertain” or “More context needed”. Each route step is labelled “Reported step” or “Uncertain step”, and each meaning “Model-reported” or “Uncertain” — “reported” means commonly reported, not verified here. An “Uncertainty notes” section lists the specific gaps for that reading, and the notice under the headword says the claims are not independently verified.

What the tool doesn't do

The tool does not look words up in a dictionary for you, and it does not prove that a route happened. The drawn path connects language forms, not places. It cannot weigh rival accounts, read dated texts or decide between dictionaries that disagree — the checking steps under each reading are how that work gets done.

How the examples and guides are made

The ten word pages are curated summaries: each follows named dictionary entries — Merriam-Webster, the Online Etymology Dictionary and the American Heritage Dictionary — and keeps first-use dates scoped to their source and sense. The ten guides are researched editorial pages; each lists its sources at the end with a “Sources consulted” date. Hero artwork and teaching figures are original illustrations labelled as such — they are decorative or explanatory, not historical evidence.

Corrections

If a generated reading looks wrong, or a guide or word page contains an error, report a problem. Reports are read, and guide and word pages are corrected when an error is confirmed; a corrected guide's “Updated” and “Sources consulted” dates are refreshed to match the change.

Word Says Hi is operated by Qomza, LLC. More about the tool is on the About page.