Meet Tari.
Tari is the dictionary’s assistant. It helps you find the right word and check any piece of writing against the 29 reviewed entries. Every suggestion links back to a reviewed dictionary entry.
Nothing here autopublishes. Tari is a starting point; the dictionary is the source of truth.
Ask Tari.
Type a question, a phrase, or just a word. Tari searches the dictionary's entries and surfaces matches.
- Works on natural-language questions, not just keywords.
- Falls back to the closest theme if nothing matches.
- Every result links to a reviewed dictionary entry.
Check a paragraph.
Paste any paragraph. Tari scans it against the dictionary and flags terms that need rethinking, with alternatives sourced from the entry.
- Highlights flagged terms in place.
- Lists suggested replacements with editor context.
- Useful for editors, comms teams, and educators.
How Tari works.
Plain-English. No black box.
Grounded in reviewed entries
Tari answers from the entries the editorial board has already reviewed and links back to them. It will not invent statistics or pull facts from outside the dictionary.
Fixed prompt, limited reach
Tari's instructions are fixed on the server and your text is handled as a topic, not as commands. Tari can't publish, edit entries, or run actions, so a hidden instruction has nothing to take over.
Cost-capped
Tari's spend is structurally capped. If a flood of queries hits, Tari auto-rate-limits; the dictionary stays up.
Privacy-clean
Your queries are never used to train any provider's models. The dictionary's commercial API tier excludes training opt-in.
Tari can't add new entries.
Only the editorial team can.
If the dictionary doesn't know a word, send it in. An editor reads every submission.