Jarvis AI Assistant command center on a Windows desktop, showing tools, voice and model routing.
Features

AI Assistant Features for Windows PC
— Tools, Voice, Memory & Routing

Jarvis AI Assistant is an AI assistant for PC that finishes tasks instead of describing them. Native Windows 10/11 app, 70+ real tools, wake-word voice (beta), memory that lasts, and your choice of 7+ AI providers.

What works out of the boxCore featuresCapability matrixPower features
Download for WindowsSee plans
What makes it different

An assistant that does the work, not just the talking.

Most “AI software for Windows” is a chat window wrapped around a single cloud model. You paste text in, you get text out, and everything that happens next — opening the file, running the command, sending the email — is still on you. Jarvis is built the other way round. It ships as a native Windows 10 and 11 app from the Microsoft Store, and it runs more than 70 real tools on your own machine: the shell, your files, git, Docker, a browser, Jira, Gmail, Google Calendar and WhatsApp among them.

Tools are the difference between an assistant that explains and one that finishes. Ask for a summary of yesterday’s commits and Jarvis reads the git log itself. Ask what is eating memory and it queries the running containers rather than guessing. Every tool call is visible and logged, so you can always see what ran and why. When a task needs several tools, the agent chains them in a single turn: read the error, find the file, draft the fix, open the ticket.

Voice, currently in beta, uses the same agent. Say the wake word and speak a request while your hands are on the keyboard or you are across the room, and Jarvis answers out loud. It is the same agent with the same tools and memory, so a spoken “check my calendar and reply to the last WhatsApp from Sam” runs exactly as a typed one would.

Memory persists across sessions. Decisions you made in a conversation last month and the projects you are working on — all of it is stored locally and available the next time you open the app, without you re-explaining anything. You can ask what you decided, tell it to remember something, or tell it to forget.

Routing is where Jarvis is genuinely unusual. It speaks to 7+ AI providers — Claude, OpenAI, Gemini, Groq, OpenRouter, Ollama and llama.cpp, plus the Claude Code command-line tool — and picks the right one per request, or lets you override with a prefix. You bring your own keys and pay your providers directly; there is no token markup in between. Routing to Claude Code means a Claude plan you already have can do the heavy lifting while Jarvis supplies the tools, the memory and the voice.

Everything runs where you are. The desktop app, the built-in web dashboard and the terminal interface share one agent, one memory and one set of settings, so nothing you configure is stranded in a single window. An Android companion app, on Google Play, lets you reach the same assistant from your phone when you are away from the PC.

What works out of the box

Built in, needs setup, or beta.

  • Chat with any connected model

    Once one AI model is connected

    Built in

  • Files & documents

    Find, read, move and organise files

    Built in

  • Desktop & app control

    Open, switch and close apps and windows

    Built in

  • Screenshots & clipboard

    Capture the screen, read and set the clipboard

    Built in

  • Scheduling, reminders & workflows

    Routines run on your PC

    Built in

  • Memory

    Stored locally on your PC

    Built in

  • Gmail & email

    IMAP/SMTP with an app password

    Needs setup

  • Google Calendar

    Sign in with Google

    Needs setup

  • Jira

    Site URL + API token

    Needs setup

  • GitHub

    Sign in or add a token

    Needs setup

  • ClickUp

    Connect your account

    Needs setup

  • Slack

    Connect your workspace

    Needs setup

  • WhatsApp

    Needs Docker Desktop + QR pairing

    Needs setup

  • Browser automation

    One-time ~35 MB browser driver download

    Needs setup

  • Android companion remote access

    Turn on Remote Access in Jarvis + Tailscale

    Needs setup

  • Semantic memory search

    Needs an embedding model

    Needs setup

  • Voice & wake word

    Turn on in Settings → Voice

    Beta

  • Messenger & Instagram

    Needs Docker; off by default

    Beta
Inside the app

A look at Jarvis, screen by screen.

Command Center

See everything Jarvis is doing, in one place.

The home screen. Talk or type to Jarvis from the bottom bar, and watch tasks, approvals, your schedule and system health update live.

  • Talk to Jarvis, or switch the bottom bar to Text and type a request.
  • Active Tasks and Current Operations show what is running and which tools it is using.
  • Pending Approvals: anything sensitive waits for your yes before it runs.
  • System Overview, Connected Devices, Upcoming Schedule and Important Alerts at a glance.
  • Quick Actions rail for the things you do every day.
Jarvis Command Center with system overview, schedule, the AI core, active tasks, approvals and quick actions

AI Core

Choose the AI model that thinks for Jarvis.

Connect a cloud provider with your own key, or run a model fully on your PC, and pick how private each request should be.

  • Provider cards for Claude, OpenAI, Gemini, Groq, OpenRouter, Ollama, llama.cpp and the Claude Code CLI: click a card to switch.
  • Paste an API key and Jarvis checks it works before saving it.
  • Built-in llama.cpp engine: one-click install, a hardware check first, and only models your PC can run.
  • Privacy mode in the same place: Strict keeps every AI call on your PC, Guarded strips secrets before cloud calls.
Jarvis AI Core with provider keys, provider cards, mode and privacy settings

AI Voice

Talk to Jarvis, and hear it answer.

Pick the voice Jarvis speaks and listens with, set up the "Hey Jarvis" wake word, and tune it to your microphone and language.

  • Choose how Jarvis hears you: on-device Whisper (private, offline), JARVIS Cloud, Groq Whisper, OpenAI, Deepgram or ElevenLabs.
  • Choose how Jarvis speaks: local Kokoro, Online (Edge), OpenAI or ElevenLabs, in the language you pick.
  • "Hey Jarvis" wake word with adjustable sensitivity and a "Tune to my voice" calibration.
  • Hands-free mode, live transcription and dictation.
  • Microphone picker, listening language and keyboard shortcuts.
  • Recent voice performance so you can see how fast it responds.
Jarvis AI Voice settings with language, microphone and transcription engines
Core Capabilities

Not a chatbot.
An intelligence system.

Jarvis was built to handle real work at scale — hundreds of requests per session, parallel tool execution, and memory that grows smarter with every interaction.

⚡
INTELLIGENCE

Neural LLM Router

Prefix-based routing sends each query to the best model. Claude for reasoning, Gemini for vision, Groq for speed, Ollama or the built-in llama.cpp engine for local privacy. Zero config switching.

claude-* → Anthropic · gpt-* → OpenAI · gemini* → Google
🔧
POWER

Agentic Tool Registry

70+ production tools: shell execution, filesystem ops, git control, Docker management, web search, Jira automation, WhatsApp/Messenger integration, and more.

Smart tool selection — only relevant tools sent per query
🛡
SECURITY

Privacy Guard Core

STRICT mode allows local models only. GUARDED mode strips credentials before a cloud model call. API keys and tokens encrypted on your PC. A local log of every cloud model call.

GUARDED: API keys, passwords, tokens redacted
🧠
ARCHITECTURE

Think-Act-Observe Loop

The agent loop runs Think → ToolCall → Observe cycles with real-time streaming events. Budget tracking prevents infinite loops. Parallel tool execution via asyncio.gather().

ThinkingEvent → ToolCallEvent → ToolResultEvent → ResponseEvent
💾
MEMORY

Cross-Session Memory

Past conversations and facts are stored locally in SQLite by default and recalled in later sessions. PostgreSQL full-text search and a Redis hot cache are optional.

Semantic recall · No context lost between sessions
🖥
INTERFACE

Every Interface, One Brain

Rich terminal CLI, FastAPI WebSocket API, Next.js dashboard, Electron desktop shell, voice, and Flutter mobile all share the same agent, tools, and memory.

CLI · Web dashboard · Desktop shell · Voice · Mobile
Full Capability Matrix

Everything Jarvis can do,
in one place.

Four pillars, one agent. Each capability is labelled: built in, needs setup (an account, key or download), or beta.

Intelligence

  • Multi-LLM routingBuilt in7+ AI providers, prefix + auto selection
  • Vision & multimodalBuilt inimage, screenshot, and document understanding
  • Streaming reasoningBuilt intoken-level Think → Act → Observe events
  • Smart tool selectionBuilt inonly relevant tools sent — saves ~6.5k tokens
  • Cost & budget guardsBuilt inper-session token + spend caps

Automation

  • Shell & filesystemBuilt insandboxed execution with approval gates
  • Git & GitHubNeeds setuplocal git built in; GitHub repos, pull requests and changelog after sign-in
  • Docker controlNeeds setupcontainers, images, logs; needs Docker Desktop
  • Jira & workflowNeeds setuptickets, boards, sprints; needs URL + API token
  • Browser automationNeeds setupnavigate, scrape, fill; one-time ~35 MB driver download

Memory

  • Cross-session recallBuilt instored locally in SQLite by default
  • Semantic memory searchNeeds setupneeds an embedding model
  • PostgreSQL & RedisNeeds setupoptional full-text search and hot cache
  • Model call logBuilt inlocal log of which provider and model were called

Experience

  • Voice (hands-free)Betawake-word → conversation → proactive idle
  • Native desktop appBuilt inElectron + MSIX desktop shell
  • Command CenterBuilt indaily briefing, weather, notes, calendar
  • Web dashboardBuilt inNext.js real-time streaming UI
  • Proactive nudgesBuilt inJarvis surfaces what matters, unprompted
Power Features

Features you won't find
anywhere else.

Beyond chat. Beyond autocomplete. Jarvis has capabilities built for professionals who need real automation — not demo-ware.

🖥

Native Desktop App

Install from the Microsoft Store and get the full Jarvis experience without a terminal. Auto-updates and deep system integration out of the box.

Microsoft Store → one-click install → auto-update
💬

Auto-Reply Engine

Automated replies for WhatsApp, with schedules and custom templates. Needs Docker Desktop and a one-time QR pairing; Messenger support is in beta and off by default.

WhatsApp · Messenger (beta) · custom rule engine
⚡

Parallel Tool Execution

Independent tools run simultaneously via asyncio.gather(). A query calling 3 tools runs in parallel, not series — typically 3× faster with zero extra configuration.

tool_a() ∥ tool_b() ∥ tool_c() → gather()
👤

Personalized Profile

Your role, projects, preferences, and custom instructions are injected into every LLM conversation. Jarvis knows who you are and adapts every response accordingly.

profile.json → injected in every prompt (60s cache)
🎙

Voice Command Mode (beta)

Hands-free operation with Whisper speech recognition and Kokoro TTS. Say the wake word, hear spoken responses, and run the same tools by voice. Currently in beta.

Whisper STT → agent loop → Kokoro TTS
🧬

Local Memory

Jarvis remembers decisions, projects and facts you ask it to keep, in a local SQLite store on your PC. PostgreSQL full-text search and semantic search are optional add-ons.

SQLite by default · PostgreSQL optional · remember / forget

Built on a world-class stack

Enterprise-grade components, battle-tested libraries, zero bloat.

Python 3.11+FastAPIasyncioNext.js 14ElectronPostgreSQLRedisPydanticWhisperPyInstallerTailwindTypeScriptWindows DPAPIpystray
View Full Guide
Who it is for

Built for people whose day is a long list of computer chores.

Jarvis was made for developers, ops engineers and anyone who loses hours to routine work on a Windows PC. If you have ever wished a chatbot could just do the thing instead of telling you how, this is that. It is available as a limited-time $9.99 one-time payment for lifetime access through the Microsoft Store, with no recurring app subscription — the details are on the pricing page.

To get going, install Jarvis from the Microsoft Store, sign in, connect an AI model (a free Groq key is the easiest start) and try a first task, such as “Show me the files in my Downloads folder.” The usage guide walks through the essential commands, power workflows and routing prefixes in detail. If you are still weighing options, our write-up of the best AI assistants for the Windows desktop compares the field honestly, including the cases where Jarvis is not the right fit. And if privacy is the deciding factor, the security page explains exactly what stays on your machine in each mode.

Get Jarvis for WindowsRead the usage guide