🆕 Now available

LLM Host

Turn one Mac into a private AI server for your iPhone, iPad, or another Mac.

Pair any of your devices (iPhone, iPad, or another Mac) with your host Mac and use the local AI models you already have installed. No API keys, no cloud round-trip, no monthly bill.

💡

Example: big models from a small laptop

Got a Mac mini with 64GB of RAM at home and an 8GB MacBook Neo on the road? Run the big models on the Mac mini and chat with them from the MacBook Neo. No need to upgrade your laptop, no need to pay a cloud provider.

📶

LAN only

Same WiFi or Ethernet network. Simplest and fastest. No router setup.

🌐

Tailscale

Secure private network from anywhere: at the cafe, on the road, in the office. Free for personal use.

🔗

Port forward

Open one port on your router for direct access from anywhere. For advanced users.

🔐

Encrypted end-to-end

Every connection between your devices is encrypted.

🛡️

Paired devices only

Anything unpaired is turned away at the door.

☁️

Never through our servers

Direct connection between your devices. No middleman.

Read the LLM Host docs →

Let VoxyAI take action on your behalf, read and reply to emails, block senders, create mail rules, send messages, and manage your calendar. Every action is governed by your policies with a default-deny security model.

✉️

Email

Full inbox management through Apple Mail, read, reply, send, delete, and create rules, all by voice or text.

Read and reply to emails
Draft, send, and delete emails
Create mail rules to block senders

💬

Messages

Read recent conversations and send iMessages. The AI adapts tone and style to match the recipient and context.

Read recent conversations
Send iMessages
Context-aware replies

📅

Calendar

Read, create, modify, and delete calendar events. Let the AI schedule meetings, set reminders, and manage your day.

View upcoming events
Create and modify events
Schedule meetings by voice

🛡️

Default-Deny Security

Nothing executes unless you explicitly allow it. No matching policy means blocked, always. You define what is permitted.

📋

Policy Manager

Create granular policies per provider, action type, contact, domain, and time window. Built-in templates for common setups.

🔍

Encrypted Audit Log

Every action proposal and decision is recorded in an immutable, encrypted log. Enterprise logging writes automation events to plaintext audit files for corporate compliance.

📨

Smart Inbox Triage

The AI reads and categorizes your unread emails by urgency, drafts replies, and lets you approve or skip each one.

⏰

Scheduled Automations

Set up time-based triggers like a morning briefing or end-of-day summary. Pauses automatically during Focus mode.

✅

Review Before Sending

Every proposed action shows an approval window where you can review, edit any field, then approve or reject before it executes.

🔌

MCP Server Support

Connect to any Model Context Protocol server to give your AI access to external tools, databases, APIs, and services, right from the menu bar.

Local (stdio) and remote (HTTP) servers
Automatic tool discovery
Encrypted configuration storage

♾️

Infinite Customization

MCP servers let you extend VoxyAI with any capability you can imagine, from querying your company database to controlling smart home devices.

Connect to GitHub, Slack, Linear, and more
Build custom servers for your own tools
Growing ecosystem of community servers

🎛️

MCP Security Controls

Every MCP tool is governed by the same default-deny policy engine. No tool runs without your explicit permission.

Per-server and per-tool policy scoping
Time-window restrictions
Automatic secret scanning on outputs

Read the Automation Guide → | MCP Server Setup Guide →

Everything you need, without giving up control.

🔒

Secrets Scanner

Built-in protection prevents API tokens, passwords, and sensitive URLs from being shared with cloud AI providers.

LLM as Judge

A second AI model verifies responses before they reach you, catching hallucinations, fabricated APIs, and material errors so you can review or retry.

🗝️

BYO Keys + Local Models

Bring your own API keys for OpenAI, Claude, Gemini, and more, or stay fully offline with Apple Intelligence and Ollama. You choose where your data goes.

🛠️

IDE Code Assistant

Works in your favorite IDE. Ask VoxyAI to write, fix, or explain code, by voice, text, or shortcut.

Natural Language Terminal

Describe what you want to do in plain English. VoxyAI converts your words into terminal commands.

📝

Auto-Generate README

Select your project and VoxyAI writes a professional README.md file, complete with sections for installation, usage, and more.

📄

Script Generator

Describe what you need in plain language and VoxyAI generates the code in any major language, tests, and creates documentation.

Speak, Type, or Shortcut

Give VoxyAI a task however you like, voice, text, or a keyboard shortcut. Use whichever fits the moment.

🤖

AI-Powered Formatting

Rough input is transformed into polished, properly formatted text using advanced AI models.

💬

Context-Aware

VoxyAI adapts formatting based on your target app - emails, Slack messages, notes, and more get appropriate treatment.

🎚️

Adjustable Tone

Choose your communication style: formal, assertive, empathetic, concise, detailed, or humorous.

🪄

Smart Paste

Formatted text is automatically pasted into your active application. No copy-paste steps needed.

📋

Clipboard History

Searchable clipboard history with AI-powered search. Browse, filter, and pin items across all apps. Encrypted with AES-256-GCM.

🌍

Translate Any Language

Say or type something in one language and get output in another. VoxyAI translates into any language you need.

VoxyAI learns your writing style, preferences, and terminology over time, so every response feels like yours.

🧠

Learns Automatically

Picks up your email signatures, greetings, technical terms, and preferences from dictation, chat, and rewrites. No setup required.

🔒

Private by Default

All memory is encrypted on your Mac with AES-256-GCM. Privacy Mode ensures your facts are only used with local AI models, nothing is sent to the cloud.

📝

You Stay in Control

View, edit, or delete any remembered fact. Toggle individual facts on or off, or clear everything with one click from the Memory window.

💬

Multi-Source Learning

Learns from dictation, chat conversations, Writing Coach usage, and optionally your clipboard, each picking up different types of personal information.

🎯

Smart Token Budgets

Automatically adjusts how many facts are injected based on the AI model. Smaller models get fewer facts to avoid overwhelming their context window.

⚙️

Automatic Maintenance

Duplicate detection, confidence decay, and stale fact pruning keep your memory relevant. Facts you use often gain priority over older ones.

🗂️

Organized Categories

Facts are categorized as Writing Style, Email Signature, Greeting, Tech Stack, Preference, Contact Name, Terminology, or Workflow.

🔎

Searchable Memory

Browse all stored facts grouped by category, or search for specific ones by content. Edit any fact directly in the Memory window.

🏢

Enterprise Ready

Administrators can disable User Memory via MDM configuration. When disabled, no facts are learned or injected into prompts.

Learn more about User Memory →

VoxyAI Your Mac. Your AI. Your Rules.