Three privacy modes
RELAXED (the default, no redaction), GUARDED (credentials stripped before cloud calls) and STRICT (local AI models only). You choose in Settings — Jarvis never decides for you.
RELAXED · GUARDED · STRICT
Security & PrivacyA Private, Local-First
A Private, Local-First
AI Assistant for Windows
Jarvis AI Assistant keeps your conversations and memory on your machine. Three privacy modes — STRICT, GUARDED and RELAXED — decide what, if anything, reaches a cloud AI model, and model calls are logged on your own disk.
Three modes, one rule
You decide where the thinking happens.
Searching for a private AI assistant usually ends in a trade-off. Either you get a capable cloud model and accept that your prompts, files and secrets travel to someone else’s servers, or you get a local model and accept that it cannot do much. Jarvis AI Assistant is built to refuse that trade-off. It is a Windows desktop app that runs on your own PC, stores its memory on your own disk, and lets you decide whether cloud AI models are used at all.
That control lives in three privacy modes, switchable at any time. A new install starts in RELAXED mode; switch to GUARDED or STRICT in the app’s settings when your work calls for it.
STRICT — local AI models only
Requests to cloud AI models are blocked, so every answer comes from a model running on your hardware through the built-in llama.cpp engine or Ollama. Your prompts are never sent to a cloud AI provider. STRICT governs AI models only: other online tools, such as web search, still reach the internet when you use them. The trade-off is honest — local models are smaller than frontier cloud models — but for a large share of everyday tasks they are more than enough.
GUARDED — cloud with redaction (recommended)
Cloud providers are allowed, but before a request goes to a cloud AI model, credential-like strings — API keys, passwords and tokens — are stripped out. Email addresses and phone numbers are not redacted. You get the quality of frontier models without handing over the secrets that tend to live in a developer’s terminal and files.
RELAXED — the default, no redaction
The mode a new install starts in. Requests go to the cloud providers you enabled as they are, with no redaction. It suits non-sensitive work; switching to GUARDED or STRICT is a single setting.
Beyond the modes, the architecture itself is local-first. Conversation history, long-term memory, settings and logs are stored on your machine, in your Windows user profile, protected by your Windows account permissions — they are not encrypted at rest. Your API keys and integration tokens are encrypted on your PC with Windows DPAPI. You bring your own keys for whatever providers you enable, so your model traffic goes straight from your PC to the provider, with no usage-based markup that gives anyone an incentive to log it. Our account server receives account, purchase and error-report data, never your conversations — the privacy policy lists exactly what.
A local log records which provider and model were called and how many messages were sent. It does not record the content of those messages.
“Private” does not mean isolated, and it is worth being precise. If you choose a cloud provider in GUARDED mode, that provider receives the redacted prompt and processes it under its own terms; Jarvis cannot change those. What it can do is make sure the choice is yours, and that a record of the call exists on your disk rather than only on someone else’s.
Security & Privacy
Your data. Your machine. Your rules.
Jarvis keeps your conversations and memory on your PC. Use local AI models only, strip credentials before cloud calls, and keep a local log of which models were called.
Encrypted API keys
API keys and integration tokens are encrypted on your PC with Windows DPAPI. Conversations, memory, settings and logs are stored locally but are not encrypted at rest.
Windows DPAPI
Local model call log
A local log records which cloud provider and model were called and how many messages were sent — not their content.
Stored on your PC
Automatic redaction
In GUARDED mode, credential-like strings such as API keys, passwords and tokens are stripped before a request goes to a cloud AI model. Email addresses and phone numbers are not redacted.
GUARDED mode
Local model support
Run AI models on your own hardware with the built-in llama.cpp engine (one-click install, no extra software) or with Ollama. In STRICT mode prompts are never sent to a cloud AI provider; online tools such as web search still work.
llama.cpp · Ollama
Stored on your PC
Conversations, memory, settings and logs live in the .jarvis folder in your Windows user profile, not on our servers.
~/.jarvis
Privacy and control
You decide where Jarvis thinks.
Jarvis makes the local and cloud boundary visible. Choose a privacy mode for the work in front of you, see which cloud models were called, and disconnect services whenever you want.
- API keys and integration tokens are encrypted on your device.
- Sensitive actions can require your approval.
- A local log records which cloud models were called, not what was sent.
Strict
Local only
Only local models (Ollama or the built-in llama.cpp engine) answer, so prompts are never sent to a cloud AI provider. Online tools such as web search still work.
Guarded
Cloud with protection
Use cloud models after credential-like strings such as API keys, passwords and tokens are stripped.
Relaxed
Direct cloud access (default)
Use enabled providers without redaction for trusted, non-sensitive work. This is the default.
Why it matters
Local-first matters more when the assistant can run tools.
An assistant that only chats can leak what you type. An assistant that runs tools — reads files, queries git, opens your inbox — can leak far more, because the tool output flows back into the model’s context. That is exactly why Jarvis was designed around privacy modes from the start rather than adding a toggle later. The more an assistant can do on your machine, the more it matters where its reasoning happens. Our features page covers everything those tools can do.
Jarvis is available as a limited-time $9.99 one-time payment for lifetime access through the Microsoft Store, with no recurring app subscription — see the pricing page. Every privacy mode is included; privacy is not an upsell. Download it from the Microsoft Store, switch to STRICT or GUARDED in settings if you need them, and read the privacy section of the usage guide for the exact setup. For a broader view of how private the alternatives really are, our comparison of AI assistants for the Windows desktop looks at what each one sends off-machine.