NovaAI

INDEPENDENT

Starting NovaAI…

Tutorial
NovaAI Ready

PRIVATE • INDEPENDENT • YOUR INTERFACE

What can NovaAI help you with?

Ask naturally. NovaAI uses the best available verified capability for each task and avoids guessing when a reliable answer cannot be established.

NovaAI starts automaticallyThe first launch may download the selected browser model. Later visits can reuse browser-cached model files where supported.

CalculatorFile readerCSV analysisPDF / document exportPrivate memoryNovaPlay handoffLive web search*Webpage reader*PDF readerXLSX readerImage + screenshot understandingVoice

NOVA CONTROL

NovaAI settings

NovaNatural improves flow, rhythm, tone and specificity. It is not designed to evade AI/authorship detectors.

👍/👎 feedback is stored only in this browser as a private improvement signal. It does not retrain or replace the live model automatically.

Local data & history

Manage information saved by NovaAI in this browser.

How this version works

NovaAI uses a browser-native open-weight model plus deterministic/browser-safe tools for calculations, supported file reading, downloadable documents and persistent memory. RD NovaSphere owns this interface, product behaviour and orchestration. The underlying base model is an open-weight model, not a foundation model trained by RD NovaSphere from scratch. Live webpage reading uses a Netlify Function. Live general web search uses replaceable zero-cost server-side discovery routes in this build; direct webpage reading uses a Netlify Function. No paid search provider is required by the baseline. PDF text and spreadsheets are parsed on the device where supported. Image understanding combines on-device OCR with an optional lightweight scene-caption model, then Nova Core reasons over the extracted visual evidence. Voice input depends on browser support; voice output uses the device speech engine.