ARCHITECTURE

Systems & Tools

Zyra AI is built as 12 systems, all reached through the live model's tool calls and routed by a single ToolRegistry. Below is every system and the exact tools it exposes — 69+ voice-callable tools in all.

12 systems69+ toolsGemini Live core

🎙️

Voice Core (Gemini Live)

ForegroundVoiceService · GeminiWebSocketClient

2 tools

The brain: a foreground mic service that streams live audio to the Gemini Live API over WebSocket, talks back in real time with barge-in, and treats the model's tool calls as commands. Every other system below is reached through the tools this core dispatches.

get_datetime

Read the phone's real local date & time (never guessed).

switch_assistant

Switch between the Zyra and JARVIS personas (Premium/Custom).

🧭

Command & Decision Engine

IntentAnalyzer → DynamicDecisionEngine

1 tool

Parses every command (voice or typed) into an intent — English + Hinglish — with no hardcoded scripts, then dispatches to the right app, action or automation path. The `control_device` tool funnels dozens of one-shot device actions through this engine.

control_device

One typed device action (action + target + value) → the canonical command pipeline.

📞

Calls & Communication

MessageSender · ZyraInCallService · WhatsAppCallMonitor · NotificationStore

10 tools

Messaging and live call handling over WhatsApp/SMS and the phone's own dialer — including talking to callers on the user's behalf.

send_message

Send a WhatsApp (default) or SMS text to a contact/number.

read_sms

Read recent SMS from the inbox, optionally from one sender.

send_file

Send a photo/video/document (found by name) to a WhatsApp contact or group.

mark_chat

Mark a WhatsApp chat read or unread.

read_notifications

Read out recent phone notifications.

reply_to_notification

Reply to a recent message notification.

read_left_messages

Read messages the auto-reply assistant took while the user was busy.

manage_call

Answer / decline / hang up / speaker / hand-over the current phone or WhatsApp call.

log_call_message

Save a message a caller left during a call.

read_call_log

Recap missed calls and messages Zyra took from callers.

📱

Device & System Control

control_device actions · QuickToggle · SystemControl

14 tools

Direct control of the phone's hardware and settings. Wi-Fi/Bluetooth/Hotspot are toggled by really driving the quick-settings tiles (Android forbids apps flipping them directly).

open_app

Open any installed app by name (fuzzy-matched).

open_settings

Open a settings section (wifi, bluetooth, data, hotspot, display…).

set_brightness · auto_brightness_on/off

Set screen brightness 0–100 or toggle auto-brightness.

set_volume · volume_up/down · volume_mute

Full media-volume control.

ringer_mode

Switch normal / vibrate / silent.

flashlight_on / flashlight_off

Torch control.

dnd_on / dnd_off

Toggle Do Not Disturb.

lock_screen · screenshot · read_status

Lock the phone, capture a screenshot, read device status.

wifi_on/off · bluetooth_on/off · hotspot_on/off

Really toggle radios via verified quick-settings tiles.

set_timer · set_alarm

Start a countdown timer or set an alarm.

read_clipboard / write_clipboard

Read or set the clipboard text.

save_contact · add_note

Save a contact or jot a quick note.

get_battery · get_network_status · get_ram

Live phone status reads (DeviceStatus).

cast_screen · remote_on/off

Cast the screen, or expose a Wi-Fi web remote to drive the phone from a browser.

🎵

Media

control_device media actions

5 tools

Search-and-play plus transport control for music and video.

youtube_play

Search & play a video on YouTube.

spotify_play

Search & play on Spotify.

music_play/pause/next/previous/stop

Control the active media session.

current_song

Say what's playing now.

youtube_stats

Look up a YouTube channel/video's stats.

🗺️

Navigation & Web

DynamicDecisionEngine · SerpApiClient

4 tools

Maps, arbitrary URLs, and live web answers. Live search runs a SerpAPI round-trip fed back into the conversation (depth-capped to avoid loops).

navigate_to

Start Maps navigation to a place.

open_url

Open any URL in the browser.

find_nearby

Find nearby places (restaurants, ATMs…).

get_weather

Current local weather (cached SerpAPI).

👁️

Screen Automation & Vision

SmartAccessibilityEngine · SmartAutomationAgent · ScreenVision (ML Kit OCR + Gemini Vision)

10 tools

Coordinate-free, dynamic UI automation: elements are found by text/intent, never fixed positions. When the accessibility tree is too sparse (games, WebViews), it falls back to visual understanding and raw taps.

read_screen

Read on-screen text via the accessibility tree.

see_screen

Screenshot and understand the screen VISUALLY (images, games, charts).

observe_screen

List visible elements with tap coordinates before acting.

act_on_screen

Run semicolon-separated steps: click / type / scroll / swipe / back / home / wait.

tap_at

Tap exact pixel coordinates where no label exists.

long_press

Long-press an element by its visible label.

scroll_up / scroll_down · swipe_left / swipe_right

Scroll and swipe gestures.

press_back · press_home · open_recents · clear_recents

System navigation, incl. verified clear-all of recents.

open_notifications · open_quick_settings

Pull down the shade or quick-settings panel.

look_camera

Take a photo and describe/solve what the camera sees.

🗂️

Files

control_device file actions

4 tools

Find and manage files on the phone by name.

find_file

Locate a file by name.

open_file

Open a file.

share_file

Share a file.

delete_file

Delete a file.

📄

Documents & Code

DocumentGenerator · CodeArtifact · WebsiteAgent (Grok via OpenRouter)

3 tools

Generate real Office/PDF files on-device, and build & live-run code and full websites from a plain-language brief.

create_document

Build a real PowerPoint / Excel / Word / PDF and open it.

write_code

Open Code Studio with generated code; runs websites/games live.

build_website

Powerful coding agent: build any website/web app/game from a brief.

🎓

Teacher Mode & Exams

WhiteboardActivity · ExamManager · ScanSolveActivity · Gemini Vision

5 tools

An interactive teaching system: draw solutions on a full-screen whiteboard, read & grade the student's own handwriting, and conduct pinned, timed exams with saved reports.

scan_problem

Open the camera to photograph a problem; OCR'd text comes back to solve.

show_whiteboard

Draw steps/diagrams (or fetch a labeled textbook image) on the board.

read_whiteboard

Read the board with vision and grade the student's markings.

exam_mode

Start/stop a focused exam — pins the phone, timer, questions.

save_exam_result

Save the score, feedback and a snapshot of the answers.

🧠

Memory, Diary, Places & Personal Data

MemoryManager · DiaryManager · PlaceAwareness · Health Connect · CalendarReader

6 tools

Long-term personal context: facts, a nightly diary, saved locations, health and calendar — injected into the persona and readable on demand.

save_memory

Persist a long-term fact about the user.

get_diary

Read Zyra's diary of the user's recent days.

manage_places

Save / list / remove places, or answer 'where am I'.

get_health

Today's steps, last night's sleep, latest heart rate (Health Connect).

get_calendar_events

Upcoming calendar events within N hours.

set_reminder

Schedule a spoken/notified reminder.

🤖

Autonomy & Automation

MissionExecutor · MissionScheduler · RuleEngine · RoutineManager · DemoRecorder

5 tools

The autonomous layer: a planner chains multiple tools to complete goals, on a schedule or triggered by events. Users can also teach shortcuts and record demonstrations as replayable routines.

schedule_mission

Schedule a recurring autonomous task Zyra runs by herself (add/list/cancel).

manage_rules

Create automation rules: 'when X happens, do Y' (time, notification, arrive/leave, charger, battery, unlock, call_from).

run_routine

Run a named multi-step routine (e.g. 'morning').

record_demo

Learn by watching: record the user demonstrating steps as a routine.

save_shortcut

Teach a voice shortcut: trigger phrase → what to do.

Tools are plan-gated — Basic = talk only, Star = +open apps, Premium/Custom = all.

See Plans Back to Tool Suites

TALK TO US

Chat with us on WhatsApp

Any questions about plans, licenses or setup? Message us directly.

Chat on WhatsApp