✦  ZenonAI · Open-Source AI Agent Platform  ✦
Open-source · 100% local · free forever

Your AI runs
on your machine.

Samantha is an open-source voice assistant. Runs offline. No cloud. No API keys. Speaks, listens, sees the world and controls your desktop.

DownloadDClone sourceG

Available for macOS, Linux, and Windows

8
LLM Models
4×
Faster STT
100%
Local & Private
0
API Keys Needed
scroll

Built by using world-class software companies

Crafted with modern tools and technologies used by leading teams.

Amazon
Anthropic
Apple
Cloudflare
DeepSeek
Docker
GitHub
Google

Capabilities

Everything Samantha
can do for you.

Six core capabilities. One voice command. Zero cloud dependency.

100% Local
⬡

Ollama Intelligence

deepseek-r1, Mistral, LLaMA 3, Phi-3. Zero subscriptions — no data leaves your machine. Swap models with one line in .env.

Low latency
◎

Low-latency Voice

Rust audio engine captures and segments speech via WebRTC VAD. faster-whisper (CTranslate2 INT8) transcribes in ~350ms locally.

Playwright
◈

Browser Automation

Browse URLs, Google search, compose Gmail — driven by a headless Chromium instance. Samantha sees your browser exactly like you do.

Cross-platform
◇

Desktop Control

Open apps, send hotkeys, take screenshots. Full OS integration on macOS, Linux, and Windows via native accessibility APIs.

Vision_io
◉

Vision

Access you your camera, capture real world persistently, analyze environment with computer vision(cv2)

SQLite
▣

Persistent Memory

Conversation history stored locally via SQLite. Samantha remembers context across every session — building richer personal interactions over time.

Live Demo

See Samantha in action.

samantha — zenonai.net

What people are saying

Trusted by builders everywhere.

“I've been using Samantha for months and never seen technology like this. Feels like talking to a real assistant.”

A
Aryan K.
ML Engineer, Berlin

“Finally an AI that respects my privacy. Everything runs locally — no subscriptions, no data harvesting.”

N
Nia O.
Security Researcher

“I replaced three paid tools with Samantha. Browser automation, email, Wikipedia — one voice command.”

T
Théo M.
Product Designer, Paris

“The memory feature is what hooked me. Remembered context from three days ago and used it perfectly.”

L
Lena V.
Developer Advocate

“Running LLaMA 3 locally through Samantha is wild. Zero cloud latency. Switched from GPT-4.”

J
James P.
Backend Engineer

“Full async, Playwright, voice I/O — it just works. The open-source agent the community needed.”

P
Priya S.
AI Researcher

“Set it up with Mistral in 10 min. Said 'open Spotify' and it just did. Still get goosebumps.”

C
Carlos R.
Indie Developer

“Voice recognition completely offline at this accuracy is unreal. This is the future of desktop AI.”

M
Mira D.
CTO, Fintech Startup

“I've been using Samantha for months and never seen technology like this. Feels like talking to a real assistant.”

A
Aryan K.
ML Engineer, Berlin

“Finally an AI that respects my privacy. Everything runs locally — no subscriptions, no data harvesting.”

N
Nia O.
Security Researcher

“I replaced three paid tools with Samantha. Browser automation, email, Wikipedia — one voice command.”

T
Théo M.
Product Designer, Paris

“The memory feature is what hooked me. Remembered context from three days ago and used it perfectly.”

L
Lena V.
Developer Advocate

“Running LLaMA 3 locally through Samantha is wild. Zero cloud latency. Switched from GPT-4.”

J
James P.
Backend Engineer

“Full async, Playwright, voice I/O — it just works. The open-source agent the community needed.”

P
Priya S.
AI Researcher

“Set it up with Mistral in 10 min. Said 'open Spotify' and it just did. Still get goosebumps.”

C
Carlos R.
Indie Developer

“Voice recognition completely offline at this accuracy is unreal. This is the future of desktop AI.”

M
Mira D.
CTO, Fintech Startup

“I've been using Samantha for months and never seen technology like this. Feels like talking to a real assistant.”

A
Aryan K.
ML Engineer, Berlin

“Finally an AI that respects my privacy. Everything runs locally — no subscriptions, no data harvesting.”

N
Nia O.
Security Researcher

“I replaced three paid tools with Samantha. Browser automation, email, Wikipedia — one voice command.”

T
Théo M.
Product Designer, Paris

“The memory feature is what hooked me. Remembered context from three days ago and used it perfectly.”

L
Lena V.
Developer Advocate

“Running LLaMA 3 locally through Samantha is wild. Zero cloud latency. Switched from GPT-4.”

J
James P.
Backend Engineer

“Full async, Playwright, voice I/O — it just works. The open-source agent the community needed.”

P
Priya S.
AI Researcher

“Set it up with Mistral in 10 min. Said 'open Spotify' and it just did. Still get goosebumps.”

C
Carlos R.
Indie Developer

“Voice recognition completely offline at this accuracy is unreal. This is the future of desktop AI.”

M
Mira D.
CTO, Fintech Startup

“I've been using Samantha for months and never seen technology like this. Feels like talking to a real assistant.”

A
Aryan K.
ML Engineer, Berlin

“Finally an AI that respects my privacy. Everything runs locally — no subscriptions, no data harvesting.”

N
Nia O.
Security Researcher

“I replaced three paid tools with Samantha. Browser automation, email, Wikipedia — one voice command.”

T
Théo M.
Product Designer, Paris

“The memory feature is what hooked me. Remembered context from three days ago and used it perfectly.”

L
Lena V.
Developer Advocate

“Running LLaMA 3 locally through Samantha is wild. Zero cloud latency. Switched from GPT-4.”

J
James P.
Backend Engineer

“Full async, Playwright, voice I/O — it just works. The open-source agent the community needed.”

P
Priya S.
AI Researcher

“Set it up with Mistral in 10 min. Said 'open Spotify' and it just did. Still get goosebumps.”

C
Carlos R.
Indie Developer

“Voice recognition completely offline at this accuracy is unreal. This is the future of desktop AI.”

M
Mira D.
CTO, Fintech Startup

Model Selection

Choose your intelligence.

Switch models with one line in .env — no code changes.

DownloadD

Latest release · AGPL-3.0 license

View source on GitHub

What is our purpose?

Our goal is to build the best multifunctional AI agent — one that can see, hear, speak, and interact with the real world.

We are working toward a seamless integration between software and hardware, combining artificial intelligence with real-world environments.

By merging virtual experiences with physical reality, we aim to create systems that are not just reactive, but also a step toward the next generation of human-computer interaction.