On-premises AI · Built, installed and supported in Seattle
Your files never leave your building. No cloud service, no remote access, installed by our technician from signed, verified software.
One server with 1, 2 or 4 RTX PRO 6000 GPUs. Chat and documents, coding help, and an image and video studio for everyone.
For your deskBuilt to order, burn-in tested for 48 hours, set up with your software and installed at your desk.
Quote only. Every GPU order includes a short US end-use form.
Security
Team AI
A GPU server in your office runs the whole suite. Each person signs in to the same workspace with their company account. Installed on-premises at your office by our technician, from a signed, verified software package.
Ask questions, draft and summarize, and search your own files with a private language model.
Help inside the editor your developers already use: completions, explanations and reviews.
Generate and edit images and short video. Jobs go into one shared queue, so everyone gets a turn.
2× RTX PRO 6000
192 GB of GPU memory in one server
Two GPUs. Room to keep chat quick for a mid-size team while longer jobs run on the second card.
Moved up one size so image and video work has its own GPU time.
Quoted for your teamInstallation, updates and support: included in your quote.
Get a quoteSuites
Every suite includes the core: private chat with company sign-in, search across your own documents, meeting transcripts and minutes, email and report drafting, and a usage dashboard by team. Installed on-premises, at your office, by our technician.
For law firms, 5 to 50 lawyers
Ask questions across case files. Summarize depositions and transcripts. Review contracts against your firm's playbook. First drafts of letters, motions and discovery responses.
For engineering, architecture and construction
Search years of project reports and drawings. Draft specs and reports. Look up codes and standards. Concept images, plus rendering and simulation on the same GPUs.
For video, marketing and creative teams
Edit by transcript. Social clips with captions. Subtitles and dubbing. Search your footage. Image and video generation in your brand style. Faster exports in Resolve and Premiere.
Pull data from W-2s, 1099s, K-1s and statements. Summarize workpapers. Draft replies to IRS and state notices.
Dictation to visit notes. Insurance and prior-authorization letters. Patient instructions in other languages. We sign a BAA.
Search specs, drawings and quality documents. Draft inspection and corrective-action reports. Can run fully offline.
Private coding assistant in VS Code and JetBrains. Ask your codebase. Code review and tests. An internal AI service for your own products.
Pre-call briefs and follow-ups. Proposals and security questionnaires. Weekly reports from your data. Invoice and form intake.
Add-ons for any suite: training on your own writing style and forms, and connections to your tools (CRM, Drive or SharePoint, Slack or Teams, e-signature).
Dashboards and support
You get a working system, not a box to babysit. Health dashboards run on your own network, for your IT team.
No alerts today. Last backup finished at 1:12 am.
Workstations
Custom builds for the person who runs models, renders or trains all day. Built to order, tested, set up with your software, then delivered and installed by us.
For one person running large models all day.
For bigger models, fine-tuning and long renders.
For image, video and 3D work at one desk.
Quiet and desk-sized. For dev teams, and for running very large open models privately.
Apple Mac Studio, M5 Ultra with up to 512 GB of unified memory, set up on-premises with our private AI suite.
We help you choose the right model.
Compare
Three ways to run private AI on your own hardware. We help you choose during the first review, based on who will use it and for what.
| Platform | Best for |
|---|---|
| RTX PRO 6000 buildsWorkstations and team servers | Many users at once, image and video generation, and training |
| DGX Spark1, 2 or 4 units | A compact CUDA box for dev teams, and clusters for very large models |
| Mac StudioM5 Ultra, up to 512 GB | A silent single box for one person or a small team |
GPUs
We also source single cards for your own builds and servers. Ordering is by quote only, and each order includes the US end-use form.
| GPU | Memory | Price |
|---|---|---|
| RTX PRO 6000 BlackwellWorkstation Edition | 96 GB | By quote |
| RTX PRO 6000 BlackwellMax-Q | 96 GB | By quote |
| RTX PRO 5000Blackwell | 72 GB | By quote |
| RTX 5090GeForce | 32 GB | By quote |
How it works
The same four steps for a team server or a single workstation. You deal with the same people the whole way through.
We talk through who will use it and for what, then size the setup and send a written quote.
We build it here and run it under load before it ships. Every GPU order includes the US end-use form.
Our technician installs it on-premises at your office from a signed, verified software package, connects your company sign-in and walks your team through it.
Updates, backups, a monthly report and health dashboards on your own network. Support is on site, or by a screen-share you start.
Done properly
High-end NVIDIA GPUs are export-controlled by the US government. We follow the rules on every order and keep the paperwork short and clear for you.
One short page: who will use the hardware and where it will be installed. It takes a few minutes.
We deliver and install around Seattle and the Eastside, and you can meet the people who built your system.
An itemized invoice that lists every part, with serial numbers for warranty claims.
FAQ
It stays on the hardware in your office. The models run on the server or workstation we install on your network. We don't keep remote access to your systems.
Updates, backups, health dashboards and alerts by text or email for your IT team, a monthly usage report, and local support on site or by a screen-share you start. It is included in your quote, not billed as a separate plan.
We confirm the build time in your quote; it depends on parts availability.
Warranty terms are spelled out in your quote.
High-end NVIDIA GPUs are under US export controls. The form records the end user and where the hardware will be installed. We sell to end users in the US only, not for resale or export.
Request a quote
A few lines is enough: how many people, what they work on, and where it will be installed. We reply within one business day, then send a recommended setup and a written quote.
Prefer to see it first? Book a demo at your office. Our technician brings a demo machine and shows it on your own sample documents.
Prefer email? Write to info@seattleworkstation.com.