Revenue model. Free tier: 20 artifacts/month, text-only capture, no integrations. Premium: $12/month or $108/year — unlimited artifacts, voice mode, calendar + email + Drive integrations, daily driving brief, PDF export, shared workspaces. Target a 6% free-to-paid conversion at scale. At 100,000 free users that's 6,000 paid × $12 = $72K MRR / $864K ARR from a single-page web app with no app-store cut. A second tier, Family ($20/mo), syncs calendars and shared lists across two adults — the realistic upgrade path for any user who has ever forgotten to tell their spouse about a school early-dismissal. CAC stays sub-$8 because the share moment is the acquisition channel.
The viral share moment. Every artifact has a "made by my assistant in 11 seconds" footer with a one-tap share. The shareable unit is not the app — it's the output. A founder posts the weekly summary their assistant produced. A parent texts the packing list. A team lead drops the meeting notes in Slack. Each share is an in-context advertisement: the recipient sees a useful, specific, formatted artifact and the small footer line, and the install is one tap because there is nothing to install — it's a URL. The mechanic mirrors how Calendly grew: every meeting booked was a billboard. Here every email drafted, every list shared, every PDF exported is the billboard.
Where it most likely fails. Three risks, in order of severity. (1) Trust collapse on a single bad artifact — if ASSISTANT mis-schedules a dentist, mis-attributes an action item, or sends an email with a wrong date, the user fires it permanently. The mitigation is a hard "review before send" gate on any outbound action and clearly-labeled confidence on every extracted entity. (2) Voice-mode coverage gaps — Web Speech API is uneven across iOS Safari versions; if the headline demo glitches on a busy parent's first try, they don't come back. Mitigation: text input is always at least as fast, and voice is positioned as the upgrade, not the floor. (3) Commoditization by Apple/Google — if Siri or Pixel Assistant ship comparable artifact production natively, ASSISTANT must already own the formatting taste and cross-context memory moat. Ship the artifact quality bar so high that the OS-level competitor feels like a generic Word doc next to it.