AI Engineering Workstation · Agent Inside

It Ships With Your First AI Employee

ABS Zaurion Ruby Mid-Tower · ZRE4565P-RP5000

Most workstations arrive empty. This one arrives working: Ubuntu with the OpenClaw enterprise AI agent framework pre-installed, a liquid-cooled 16-core AMD EPYC 4565P, and an RTX PRO 5000 Blackwell with 48 GB of ECC GDDR7 — sized exactly for the mid-size LLMs that real products run on.

Illustrative scene — an engineer watching an AI agent complete its task list unattended
OpenClawagent framework, pre-installed
48 GBGDDR7 with ECC
16-coreEPYC 4565P · up to 5.7 GHz
128 GBDDR5 ECC UDIMM
2× 2 TBM.2 NVMe
1000 W80 PLUS Platinum
Day One, Not Month One

From Box to Working Agents

The point of pre-installed software is time you don't spend. Here's the realistic ramp:

Hour 0 · Unbox

Boot straight into Ubuntu

No OS install, no driver hunt — CUDA-ready Ubuntu is preloaded and stress-tested before the box ships from the USA.

Hour 1 · Connect

OpenClaw is already there

The enterprise AI agent framework is pre-installed. Point it at your ticket queue, docs, or repos and define your first automation — framework setup time: zero.

Day 1 · Serve

Load a mid-size model into 48 GB

ECC GDDR7 holds the mid-size open-weight LLMs that production tools actually use. Want it fully turnkey? An optional pre-engineered AI environment with LLM, RAG, and automation is available by quote.

Week 1 · Automate

Agents in production

Ticket triage, report drafting, document processing — running locally on your hardware, with fixed costs and nothing leaving the building.

The Hardware Behind the Agent

Small Tower. Server Manners.

CPU
AMD EPYC 4565P — 16-core, liquid-cooled, 4.3 GHz base / 5.7 GHz max turbo, 64 MB L3 (Socket AM5)
GPU
NVIDIA RTX PRO 5000 Blackwell — 48 GB GDDR7 with ECC
Memory
128 GB DDR5 ECC UDIMM — expandable to 256 GB across 4 DIMM slots
Storage
2× 2 TB M.2 NVMe — OS and data kept apart, 4 TB total
Network
1× 10 Gb/s + 2.5 Gb/s LAN + dedicated 10/100/1000 Mbps management port
Expansion
PCIe 5.0 x16 + PCIe 4.0 x16 on the ASUS Pro WS B850M-ACE SE board
Power
1000 W 80 PLUS Platinum — premium efficiency, sized for full-load inference
OS
Ubuntu pre-installed (Windows available by quote)
Same Listing, Three Weight Classes

Pick the VRAM Your Models Need

This listing configures from starter to monster — same agent-ready Ubuntu setup on every rung (an AMD Ryzen 9950X3D CPU alternative is also offered; compare options on this page):

Starter

RTX PRO 4500 Blackwell

32 GB

Prototyping, smaller open-weight models, and agent workloads that lean on APIs and retrieval more than raw VRAM.

This Config
The Engineering Sweet Spot

RTX PRO 5000 Blackwell

48 GB

The listing's own positioning: testing and deploying mid-size LLMs — with ECC on the GPU memory for production trust.

Step Up

RTX PRO 6000 Blackwell

96 GB

For the largest single-GPU models this family offers — also available in a MaxQ variant on this listing.

What Teams Run On It

Four Ways It Earns Its Desk Space

Illustrative scene — engineer watching an AI agent execute automated tasks in a terminal
01

Agents That Work While You Don't

OpenClaw ships pre-installed: build agents that triage tickets, draft replies, process documents, and chain multi-step workflows — on your hardware, under your policies.

Ticket triage · report drafting · workflow automation
02

Test & Deploy Mid-Size LLMs

48 GB of ECC GDDR7 is the honest sweet spot for the open-weight models production tools actually use — evaluate candidates side by side, then serve the winner privately.

Model evals · private serving · fixed costs, no per-token fees
Illustrative scene — ML engineer comparing model outputs with an evaluation scorecard
Illustrative scene — startup team reviewing an internal AI dashboard together
03

Internal Tools Your Team Actually Uses

RAG over your docs, search over your tickets, automations over your processes. Want it turnkey? The optional pre-engineered AI environment bundles LLM, RAG, and automation — by quote.

Company knowledge base · document Q&A · process automation
04

Serious Compute, Office-Friendly Body

A liquid-cooled 16-core EPYC in a mid-tower that fits under a desk — with a dedicated management LAN so IT can run it like server gear anyway.

17.5″ × 9.17″ × 19.45″ · liquid-cooled · out-of-band management
Illustrative scene — compact mid-tower on a tidy desk beside a single monitor

Scene photos are AI-generated illustrations of typical deployments, not photographs of this exact product. Agent session shown in the hero is a stylized illustration.

FAQ

Quick Answers Before You Buy

What does "OpenClaw pre-installed" actually mean?

The OpenClaw enterprise AI agent framework comes installed and ready on Ubuntu — build and run agentic workflows the day it arrives, with no framework setup.

Which models can it run locally?

The RTX PRO 5000 Blackwell's 48 GB of ECC GDDR7 targets mid-size open-weight LLMs — exactly the listing's stated positioning. A pre-engineered LLM/RAG/automation environment is optional, by quote.

Can I get it with Windows?

It ships with Ubuntu pre-installed; Windows is available — request a quote from the listing.

What if I need more GPU later?

Sibling configs on this listing go up to the 96 GB RTX PRO 6000 class, with an AMD Ryzen 9950X3D CPU alternative too. Memory expands to 256 GB across the 4 DIMM slots.

Is it eligible for government / public-sector purchasing?

It is TAA compliant and assembled in the USA, with compliance documents available for purchase review.

How is it supported?

1-year parts and labor limited warranty backed by ABS manufacturer support. Build to Order — custom built and stress-tested after purchase.

OpenClaw pre-installed
Assembled in USA
TAA Compliant
Liquid-cooled · ECC end to end
ABS Zaurion Ruby ZRE4565P-RP5000 · ASUS Pro WS B850M-ACE SE platform · Positioning and specifications per the Newegg listing. Optional components and software are configured at time of purchase; the pre-engineered AI environment (LLM, RAG, automation) is quoted separately.