Chat & voice

A conversational front door to the analytics, with voice in and out.

At a glance

Built-in engine: instant, offline, never invents numbers
Optional Claude API or Ollama for more natural answers (facts only)
27 regions and languages for speech input and voice, e.g. NZ, AU, UK, US, India and South Africa English, Te reo Māori, Sinhala, Tamil, Hindi, Chinese, Japanese, Korean, Arabic and more
Speed and pitch controls; Test voice button; ▶ Briefing reads today's summary aloud
Opens with a spoken daily briefing that greets you by name (Pavi, Sir, Ma'am or your own) and time of day
Signs off when you say 'done for the day', 'good night' or tap 🌙 Done

Speech input uses the browser's speech recognition in your chosen language.
Voices are grouped: best match for your region, same language, then other English voices. If a voice isn't installed on your device, Shelly says so and uses the closest one.
With the built-in engine, written answers are in English. With Claude or Ollama, answers are written in your chosen language.
On open, Shelly picks the greeting from your device clock, then reads the date, this week's sales, today's P1 actions, any recall and the decisions waiting for you.
Browsers only allow sound after you touch the page. Shelly tries to speak straight away; if the browser blocks it, the first tap anywhere (or the 'Tap to hear' button) starts the briefing.
The sign-off summarises how many actions you ticked and decisions still open, then says goodbye to suit the time of day.

Skill tag

Each answer shows which skill produced it, e.g. 'skill: retail.reorder'

Time of day

5am–noon morning · noon–5pm afternoon · 5pm–10pm evening · 10pm–5am late night

Settings

You & your day: what Shelly calls you; briefing every open, first open of the day, or off

Available voices depend on your phone or computer. Add more in its speech settings.
Voice input works in Chrome, Edge and Safari. Firefox doesn't support it yet.

What can you do?Play the briefing

Type or say these in the chat on the main page.