Questions you'd ask if
we were across a table.
What's a "personalized voice," exactly? +
Upload 8–15 samples of your own writing (emails, memos, drafts — whatever you actually produce). We train a small adapter — a LoRA — on top of the base model. The adapter learns your sentence rhythm, hedging habits, vocabulary, and switch points between languages. The adapter is yours; you can delete it; we don't train the base model on it.
Is this for evading AI detection? +
No. Parlaví makes the writing sound like you because it is you — drafted faster. If you came looking for stealth, you came to the wrong place.
Which languages? +
26 at launch — covering the Romance and Germanic families, Mandarin (simplified & traditional), Japanese, Korean, Arabic, Hindi, Turkish, Polish, and Czech among others. Code-switching between any two of them is supported within the same draft. Vietnamese and Hungarian are in late testing.
What happens to my writing? +
Drafts are not stored after the response is returned. Your personalized-voice training samples are encrypted at rest, accessible only to you, and deletable on demand. We do not train the base model on user data. The Team plan adds an enterprise DPA; the API plan adds residency commitments and an optional BAA.
How does pricing compare to alternatives? +
The closest comparable charges $36/mo for a feature set that does not include a personalized voice or real multilingual writing. We don't price-match — we price for the time you get back.
Can I switch plans? +
Yes — at any time, in one click, with prorated billing. Team-to-Individual downgrade keeps your personalized voices; the shared organization voices stay with the team. API contracts are annual but include a 30-day exit clause.