Coming soon · Launch registration is not open yet.
PebbleRack concept: a compact metal enclosure with a carry handle, visible Ethernet connections, a computer and storage bays.
System concept. Final hardware, rack design and fit are still being evaluated.

In development

The Future of AI is local

AI on hardware you own.
A home lab you help build.

Meet the idea
Front view of the proposed PebbleRack system, showing mounted components and neatly routed Ethernet cables. This complete rack is a concept, not an available kit.
Rack system concept · Not an available kit

The best part?
You built it.

Mount the pieces. Connect Ethernet and power. Get to know the little environment you’ve made your own.

That’s the experience we’re working toward: compact hardware, considered rack parts, and guidance from the first connection to your first local AI task.

First planned computer

Pebble AI Pro 128

128GB unified memory

A standalone computer. The surrounding rack, networking and storage shown here are concepts for a broader system.

What’s confirmed, and what’s still taking shape?

128GB unified memory is the planned starting point, shared by the system and graphics. Final specifications, software, price and launch date are unconfirmed. Rack fit, cooling, RAM and SSD upgradeability still need validation.

AI at home.
On your terms.

We believe you should have the choice and the ability to run AI on your own hardware.

Keep the cloud tools you love. Give suitable work a place to run locally, with more control over where it happens.

Explore 25 free home-lab guides →

Can I stop paying for tokens?

Work that runs entirely on your hardware has no cloud provider’s per-token inference charge. Hardware, electricity and some software still cost money. Model capacity and speed depend on the setup.

Can local AI help me get more from Claude or Codex?

A separate local model can handle suitable tasks alongside your cloud tools. It doesn’t add tokens to those accounts or raise their limits. Automatic integration and throughput gains are not validated PebbleRack features.

For your existing accounts, start with Claude’s usage guidance or OpenAI’s usage guidance.

Will my prompts and files stay at home?

A fully local workflow can keep them on your machine. Cloud models, web search and other connections can send data elsewhere. We plan to document the data flows of each validated setup.

Where do I start with AI at home?

Choose one task, then check whether your current computer can run a suitable local model. You don’t need a rack to begin. We’re developing a guided route to dedicated hardware; tested PebbleRack walkthroughs are still ahead.

Help shape PebbleRack

What would you want from a home AI setup?

We’re looking for people willing to spend 15 to 20 minutes talking about the equipment they’re considering, their budget, and what makes choosing it difficult. Whether you’re just getting started or already run a home lab, we’d like to hear from you.

Email us about an interview

As a thank-you for completing an interview, we’ll offer $10 in PebbleRack store credit if and when our store launches.

One credit per person. No purchase is required to take part. The credit has no cash value, and no store launch date is confirmed.

Emailing about an interview does not subscribe you to launch updates. We’ll use your message to arrange the conversation. Privacy notice

Launch registration is not open yet. Please check back soon.

Hardware, software, pricing and availability are still being evaluated.