NOMAD Pocket portable offline AI product

WHY NOMAD

AI that goes with you.
Even off the grid.

When the connection is unreliable, work still needs to get done. NOMAD Pocket is a physical USB AI product in development, designed to keep your local model, knowledge files, and saved workspace together. Plug it into another compatible Windows or Linux PC and your saved chats, notes, and memory come with you.

Check your PC first
01 / WORK OFFLINE

Keep going without a live connection.

Use the local model, knowledge, and context saved on NOMAD Pocket when getting online is difficult. New downloads and updates still need internet access.

02 / CARRY YOUR WORK

Your memory moves with you.

The planned USB workspace keeps your saved chats, notes, and reusable memory alongside models and reference files. Move NOMAD Pocket to another compatible PC and pick up where you saved your work.

03 / FIT YOUR PC

Start with the computer you already own.

Whichever compatible PC you plug into supplies the processing power; your saved workspace stays on NOMAD Pocket. Our checker helps you explore public models based on that PC's specs.

BUILT FOR INDONESIA

Good work should keep moving, even when the signal stops.

Indonesia stretches across more than 17,000 islands. For people on ships and offshore sites, teachers in remote schools, and teams working where connectivity is limited or disrupted, getting online is not always an option. We are building NOMAD Pocket to help them research, draft, and organize everyday work locally, without waiting for a connection. The USB is designed to carry local AI models, preloaded knowledge, and the workspace you save; your Windows or Linux PC supplies the processing power.

Island count: Badan Informasi Geospasial ↗

Person using a laptop during a flight
Teacher working with children in a rural classroom
Worker using a laptop near an offshore platform
Person using a laptop by the coast

PRACTICAL BY DESIGN

Made for Windows
and Linux PCs.

NOMAD POCKET / FIRST LOOK01 — 01
RESOURCESUPPORTNOTES
Computer OSWindows / LinuxNOMAD Pocket supports both; check the available setup for your OS
CPUExact Intel / AMD modelSelect your processor below; unlisted models are not assessed
Local AIModel-dependentOpen GGUF text models are suggested below via Hugging Face
GPUOptional · exact modelVRAM counts only for a listed GPU with a documented backend candidate
RAM + storageEnter your PC specsModel files, context, and runtime need extra headroom

MODEL COMPATIBILITY CHECKER

Which local AI model
fits your PC?

Select your exact Intel or AMD processor. Add a dedicated GPU model if you have one, then enter RAM, VRAM and free storage. We check popular open-licensed GGUF text models on Hugging Face and show up to five capacity estimates.

The CPU/GPU catalogue is limited and checked locally. Unlisted CPUs are not assessed; unlisted GPUs are treated as CPU-only. We fetch model listings and file sizes, never send your PC specs to Hugging Face.

CPU not listed? Send us the exact model so we can review it. Your specs are entered manually; browsers cannot reliably inspect PC hardware.
Where do I find my CPU and GPU models?

Windows: open Task Manager → Performance. Linux: check System Information, or use lscpu for the CPU and lspci for the GPU. Enter dedicated GPU VRAM, not shared system memory.

GOOD TO KNOW

Questions before you
go offline.

NOMAD Pocket is still in development. Here's what the planned USB workspace and the model checker can tell you today.

Can I use NOMAD Pocket without internet?

That is what we are building it for: local AI work with the model and knowledge files already on the USB. A connection is still needed to download new models or updates. The website's Hugging Face model checker also needs internet to fetch current listings.

Is the USB a computer on its own?

No. NOMAD Pocket is a planned physical USB product that carries the local workspace. A compatible Windows or Linux PC supplies the CPU, RAM, and optional GPU used to run a model.

Will my saved memory follow me to another PC?

That is the goal: keep the chats, notes, and reusable memory you save on NOMAD Pocket, along with its models and knowledge files. Plug the USB into another compatible Windows or Linux PC to continue from the same saved workspace. The new PC supplies the processing power. Final saving and backup details will be confirmed before launch.

Do I need a dedicated GPU?

Not always. Smaller quantized models may run on a CPU when there is enough system RAM, although speed depends on the PC and the final runtime. A supported GPU with enough dedicated VRAM may help; the checker shows capacity estimates for both paths.

How do I find a model that fits my PC?

Select your exact CPU, add your dedicated GPU and VRAM if you have one, then enter RAM and free storage. The checker looks at file sizes for popular public GGUF text models on Hugging Face. Its suggestions are estimates, not performance benchmarks or a guarantee that a model is included with NOMAD Pocket.

What if my CPU or GPU is not listed?

We will not guess about an unlisted CPU. Send its exact model to nomadpocketai@gmail.com for review. If a GPU is not in the checker catalogue, its VRAM is ignored and only CPU estimates are shown.

Can offline AI search the live web or know today's news?

No live web access means no live search or automatic news updates. Answers can use the local model and available knowledge files; current information requires a connection and updated files.

Is NOMAD Pocket available to buy?

Not yet. NOMAD Pocket is in development, and final hardware details, included files, and availability will be shared before launch. Follow @nomad.pocket on TikTok for updates.

MEET NOMAD POCKET

Bring your work.
Leave the signal behind.

NOMAD Pocket is being developed to put an offline AI workspace in a portable USB device. From smaller models for modest PCs to more capable models for stronger machines, start by finding a fit for the computer you already use.

Check your PC first →

NOMAD Pocket is coming soon. This page introduces the planned physical device; the compatibility checker estimates suitable public models and does not install them or process your AI requests.