What works today — honestly

Home-page demo

3 questions per day, then sign in to continue.

Calm Room (anxiety support)

8 replies/day as a guest, 40/day signed in. Not a replacement for therapy.

Voice input & read-aloud

Uses your browser. Works best in Chrome/Edge; Safari/Firefox voice input is limited.

Accounts (email + Google) & saved missions

Conversations are saved to your account.

Direct mode with Relay

Each model gives 8 free answers a day; when one runs out or fails, the next model picks up the full conversation and continues seamlessly.

Memory core

Facts you save are shared with every model.

Universal translator

Free on-device in recent Chrome; other browsers fall back to AI translation.

Translate a whole conversation

Originals stay one tap away.

Reply in any language

Pick the language under the message box.

Council (side-by-side models)

Daily trial of 10 runs shared across credit modes.

Supreme (fused answer)

Same daily trial.

Forge (refinement chain)

Same daily trial.

Buying credits / plans

Payments aren't set up yet, so only the daily trial exists.

Voice-first talk with a living logo

Tap the logo, speak, hear the answer. Logo reacts to listening / thinking / speaking. Voice input needs Chrome or Edge.

Research on the live web with sources

10 runs a day. Source links are written by the AI, so double-check important facts.

Vision — ask about a photo

10 photos a day. Reads text in images too.

Image studio

10 pictures a day, with live preview. Famous characters and brands may be refused.

Maps — find any place

OpenStreetMap search and map.

Reminders

Saved on your device; alerts only while the app is open.

PDF questions in chat

Attach PDFs (up to 30 MB) to any mission with the paperclip and ask about them; every model and the relay see them. Scanned PDFs without text aren't readable yet.

Memory export

Download everything the AIs know about you.

Labs: Planner, Decision simulator, Creative studio, Code lab, Research report, Tutor

20 runs a day shared. Copy or download every result.

Mission Mode

One big goal → plan you can edit → up to 7 specialist agents in parallel (with web research) → verification → downloadable delivery. 60 AI steps a day.

Business & Finance lab

Budgets, KPIs, forecasts. Never moves money.

Auto model pick

Choose 'Auto · best fit' in Direct mode — routes code, math, writing, news and quick asks to the best model at no extra cost.

Council specialist roles

Each model gets a role: reasoner, evidence, critic, technical, strategist, creative. Supreme mode fuses them.

Search past conversations

Search box in the side menu looks through titles and messages.

Voice choice & speed

In Command Center. Voices available depend on your device.

Admin system check

Runs every feature live and asks the AIs what to improve.

Control room: feature switches

Turn every feature on or off for your account.

Privacy center

Choose whether models see your memory and learn from ratings; delete all your data in one tap.

Usage dashboard

See today's use against every free allowance, plus the last 14 days.

Feedback learning

Thumbs up/down on any answer; every model adapts to your ratings.

Reminder alerts with quiet hours

Pop-ups and spoken reminders, silent in quiet hours. Only while the app is open.

Change a photo

Studio. Shares the 10/day image allowance. Keep refining the result.

AI voiceovers

Studio. 8 voices, downloadable. 15 a day.

AI video creation

Studio — off until you switch it on. ~6s clips with sound, saved to your Vision Board. 2 a day.

Code runner

Runs JavaScript and HTML in a sealed box. 'Run it' from Code Lab answers.

Interrupt mid-answer

Speak over Supreme in Talk and it stops to listen. Works best with headphones; Chrome/Edge.

Wake word “Supreme”

Switch on; listens only while the Command Center is open in Chrome/Edge.

Hologram core

3D projected S13 core in the Command Center.

Send & calendar

Send answers by text, email or any app; add reminders to any calendar. You always press send.

Acting inside your accounts (auto-send, smart home)

Needs connections to each of your accounts and your approval per action.

Hand-gesture control

Shown as a switch; browsers can't do it reliably yet.

Grok, DeepSeek, Mistral, Perplexity

Needs your own API keys from those providers.

Always-on wake word with the app closed

Phones and browsers don't allow listening when the page is closed.