Jarvis Demo (Runtime Prototype)

This is a demo build of the Jarvis Runtime Prototype.

Jarvis is designed to work with multiple AI services.
You choose which AI provider it connects to.

⚠️ Important:
Jarvis does not include a built-in AI model.
It connects to the selected AI service using your API key.
This is expected, normal, and safe.

How to run

Open the .env file

Insert your own OpenAI-compatible API key

Run JarvisDemo.exe

Open your browser and go to:
👉 http://localhost:4000/jarvis.html

What you need to get Jarvis running

Just 3 simple steps.

1️⃣ Choose an AI service

Jarvis supports popular AI providers, including:

Mistral

DeepSeek

xAI (Grok)

HuggingFace

Google Gemini

Anthropic (Claude)

👉 You only need to choose one.
👉 If you don’t like one — just switch the key.

2️⃣ Get an API key

An API key is a long string of letters and numbers.

You can get one here:

Mistral → https://console.mistral.ai

DeepSeek → https://platform.deepseek.com

HuggingFace → https://huggingface.co/settings/tokens

xAI → https://console.x.ai

Google Gemini → https://aistudio.google.com

Anthropic → https://console.anthropic.com

💡 This usually takes 1–2 minutes.

3️⃣ Insert the key into the .env file

Next to JarvisDemo.exe you’ll find a file called .env.
Open it with a regular text editor.

The file already contains examples.
You only need to uncomment one option and paste your key.

Example (Mistral):

LLM_PROVIDER=openai-compatible
LLM_API_KEY=PASTE_YOUR_KEY_HERE


Example (Google Gemini):

LLM_PROVIDER=gemini
GEMINI_API_KEY=PASTE_YOUR_KEY_HERE


👉 Nothing else needs to be changed.
👉 One key = one working Jarvis.

Start Jarvis

Run JarvisDemo.exe

Open your browser

Go to:
👉 http://localhost:4000/jarvis.html

That’s it. You can start talking 🙂

If something doesn’t work

Don’t panic — most issues are fixed in seconds.

❌ Jarvis doesn’t respond

Make sure the key was pasted without spaces

Make sure only one provider is enabled

Try another AI service (just switch the key)

❌ Access or quota error

Some providers require billing or explicit access

This is a limitation of the AI service, not Jarvis

Can I change the AI provider?

Yes.
As often as you want.

Jarvis is the runtime.
You choose the AI.

Important to know

Jarvis sends data only to the selected AI provider

Your API key is stored only on your machine

This is a demo version for evaluation and testing

In short

Download → get a key → paste it → run → talk

No registration.
No Jarvis accounts.
No unnecessary nonsense.

Provider Profiles (v1.1)

Jarvis supports provider profiles — separate .env files for each AI service.

This is intentionally designed to simplify provider switching
and eliminate confusion with keys, URLs, and settings.

How it works

Inside the providers/ directory you’ll find ready-made profiles:

Anthropic/

Google (Gemini)/

DEEPSEEK/

MISTRAL/

HUGGINGFACE/

XAI/

OpenAI-compatible/

Each profile contains its own .env file, already configured for that provider.

How to select a provider (the easiest way)

1️⃣ Open the folder of the provider you want
2️⃣ Copy the .env file
3️⃣ Paste it next to JarvisDemo.exe
4️⃣ Insert your API key
5️⃣ Run Jarvis

Done.
No switches. No restarts. No rebuilds.

Why this exists

to avoid mixing up API keys

to switch providers quickly (copy one file instead of editing many variables)

to keep demo builds stable

to isolate each provider at the configuration level

If one AI no longer works for you — just use another profile.

Jarvis stays the same.

Important

Demo builds contain no embedded API keys

All keys are stored only by you

Jarvis sends data only to the selected AI provider

In short

One provider → one .env
Copy → paste key → run