The file
A demand letter, a deposition excerpt, a client’s medical record.
Private · On-Premises · Yours
Your firm’s own AI.Nothing leaves the building.
We buy the hardware, build it, configure it around the way you already work, and hand you the keys. One server in your office — no subscription, no per-seat fees, and no vendor holding your client files.
A fifteen-minute call is enough to tell you whether this fits your firm.
- Hardware
- Build
- Configuration
- Training
- Support
Running on your network
- 5.91″Wide & deep
- 2.01″Tall
- 3.26 lbWeight
Where this starts
AI is already inside your firm
That decision has already been made — by your associates, your paralegals and your assistants. The open question is whose computer your client files sit on when it happens.
78%
of employees say they use unapproved AI tools.
WalkMe / SAP, AI in the Workplace Survey, 2025
79%
of legal professionals report using AI tools — while 44% of firms have no AI policy at all.
Clio Legal Trends, cited by the NC Bar Association, 2026
11%
of everything employees paste into AI tools is sensitive material.
Cyberhaven, 2026
Every one of those sessions runs on a server you do not own, under terms you never negotiated.
The invisible path
What happens when an employee pastes in a client document
A personal account
Free tier, personal email, no admin console, no record on your side.
Someone else’s servers
Held under consumer terms your firm never reviewed or signed.
Out of your hands
Retention you do not set, training you cannot audit, breach and discovery exposure you inherit.
No log. No copy. No answer for the client who asks where their information went.
How it works
One server, inside your perimeter
Your team signs in from any browser on the office network. The model, the documents and every chat stay on hardware in your building.
Your office network
Attorneys
Paralegals & staff
Authorized users
Your AI server
- Open WebUI — the chat interface your team signs into.
- A local model — GPT-OSS, Qwen or Gemma, selected to best fit your business.
- An admin account — full visibility of the chats and documents of every authorized user.
Everything above happens on hardware in your building.
The public internet
Not required. We can block it outright, or open only what you approve — a web-search tool, or a cloud model reachable from anywhere.
If your internet connection drops, the AI keeps working
What your team sees
It looks like the tool they are already using
A chat window with a message box. Nobody has to learn a new way of working.
Familiar on day one
A chat window with a message box — the same shape as the tools your team already opens.
Their own sign-in
Each person has an account. What they can reach is set by their role in the firm.
A shared prompt library
The firm’s best prompts written once, then available to everyone as a single click.
Answers from your own files
Load documents into the server and ask questions about them directly.
Nothing to install
Any browser on the network that you have granted access — desktop, laptop or phone, in the office.
No waiting on a vendor
The server answers at the speed of your own network, with no queue, quota or usage cap.
The part that matters
Configured around how your firm actually works
The value is in the setup — and this is the work we do with you, on site.
Groups and permissions
Partners, associates, paralegals, employees — each group sees only the models and collections it should.
Approved models only
Only the models you have vetted appear in the menu. No silent switch to something new.
Retention and records
Chat retention set to your policy, deletion locked where you need a record, complete admin visibility.
Sign-on and offboarding
Open registration when you hire, and revoke an account the day someone leaves the firm.
You get a written record of every setting we apply — in case you want to change it later.
The engagement
What we do, in order
-
Build phase
We order the AI server hardware and configure it with the model and software your firm will run.
-
Install and deploy
On-site installation, accounts and access created, with a written record of every setting we changed.
-
Train and hand over
A one-hour hands-on session for your admin covering the software and the start-up and shutdown procedures. All applicable documentation is supplied.
One firm, one box, one point of contact — us.
Straight answers
The questions we always get
Will this server be as good as ChatGPT or Claude?
For the work a firm does most — summarising, drafting and searching your own documents — it is close enough that most people cannot tell the difference. For the hardest cognitive reasoning, the frontier cloud models still have a slight edge.
What happens if it breaks?
The AI hardware we use comes with a one-year warranty from the factory, and we add one free year of technical support that runs alongside it — so for that first year, you're covered by both at the same time.
What is the lead time?
From ordering the AI computers to a fully deployed server is about three to four weeks, mostly dependent on hardware availability.
Who has access to the server?
After the server is built, configured and deployed, the only people with access are the ones you choose.
How many people can use it at once?
One person or thirty, the server costs the same — there is no per-seat license. We size the hardware to your headcount during the first call so the response times stay comfortable at your busiest hour.
Do we need an IT department to run it?
No. We hand the server over configured and running, and train one admin at your firm on the day-to-day tasks: adding a user, removing a user, and the start-up and shutdown procedures. Everything else is documented.
Talk to us
Fifteen minutes tells you whether this fits
Call and we will walk through your firm’s size, the work you would put through it, and what the hardware would cost you. No obligation, and no pressure to buy anything on the call.
720-937-6463