Lia - Give your mind some room

Lia helps people managing home, work and family capture useful thoughts by voice. Led by Ariane Figueiredo, it combines a responsive browser agenda with a concise companion powered by AssemblyAI Voice Agent API. Family feedback showed that voice entry was too hard to discover. R08 puts a microphone action and spoken example on the first screen. A setup dialog brings language, demo access and consent together; typing remains optional. Complete drafts can be reviewed and explicitly saved in one tap. A spoken task becomes a draft, never an automatically saved commitment. The application checks an exact quote against the transcript, resolves supported explicit date and time expressions locally, and leaves unknown dates for clarification. Editing, completion, search, undoable deletion, moving unfinished work to tomorrow and calendar export are available. The local date updates daily without deployment. Six interface and conversation languages, matched narrators and volume controls support accessibility. Speech and current task context are sent to AssemblyAI only after consent. API credentials remain server-side and sessions are time-limited. Tasks stay in the browser; there is no cross-device sync or closed-page notification service. Thirteen automated tests pass. A live API test with synthetic Portuguese speech produced a source-validated draft for tomorrow at 12:00. The updated presentation labels synthetic narration, actual API response audio and real UI screenshots. Physical-microphone acceptance and testing in every language remain pending. Further family testing will assess clarity, corrections and daily usefulness. No adoption or business outcomes are claimed.