VendeeX

Your personal AI shopping agent, running on your computer.

VendeeX is designed to work for you, not for advertisers, marketplaces or merchants. The AI runs locally on your own computer, and the personal information it uses stays there. Vendee Labs does not see your conversations, preferences, shopping history or personal context.

Choose your platform below to get started.

Download for macOS Windows Coming soon Linux Coming soon

Your AI runs on your computer

VendeeX runs its AI locally on your computer rather than sending your conversations to a cloud AI service.

Installation is one click. VendeeX installs the application and automatically retrieves and verifies the local AI model it needs to run.

Once installed, the AI operates on your computer. Your conversations and personal shopping context remain on your device and are not visible to Vendee Labs.

Because the AI runs locally, performance depends substantially on the capabilities of your computer.

Recommended System Requirements

macOS

Apple Silicon, M1 or later.
16 GB memory.
12 GB free storage.

Windows

For the best VendeeX experience, we recommend a recent Windows 11 PC designed for local AI workloads, ideally a current-generation AI PC with at least 16 GB RAM and modern GPU or NPU acceleration.

Older Windows PCs may still run VendeeX, but response times can be substantially slower depending on the processor, graphics hardware and available memory.

For the fastest local inference, newer systems with dedicated AI acceleration are strongly recommended.

Linux

64-bit Linux.
Modern 64-bit Intel or AMD processor.
16 GB RAM.
12 GB free storage.

Under the hood

VendeeX uses the same core local AI architecture across macOS, Windows and Linux. Platform-specific acceleration allows the application to make the best use of the hardware available on each operating system.

Desktop application: Tauri
Language model: Qwen 3.5 4B
Quantisation: Q6_K
Model format: GGUF
Inference engine: llama.cpp b9608
macOS acceleration: Metal
Windows acceleration: CUDA or Vulkan
Execution: Local on the user's device
Model delivery: Automatic, cryptographically verified local model installation
Recommended system memory: 16 GB or more