Home / Private AI
AI your staff can use, on a server you own.
Someone pastes a client file into a public chatbot to get a quick summary. Now that file sits with an outside company. Wirestack, an Orange County software company, sets up AI on hardware you own instead. Your staff get a ChatGPT-style screen and can ask questions about your own files. Prompts and documents stay on your network. Running in under a week.
Nothing sent outside the building
The staff screen: a question, and an answer that points to the document it came from.
For offices whose files can’t go into a public chatbot.
Home health and hospice
Visit notes, referrals, care plans and policies are full of patient details nobody should paste into a public chatbot. We already build software for home health agencies and know how much of their day is paperwork.
Law firms and accountants
Contracts, statements and case files your clients expect to stay put. The AI reads them where they already live.
Any office with private files
Price lists, bids, HR files, years of proposals on a shared drive. Staff ask questions in plain English and get answers that point back to the document.
One setup fee, then one monthly rate.
- A written plan first: which machine to buy, what it will run, and what we charge. Before you spend anything.
- Install and setup on the machine, on your network.
- A ChatGPT-style screen in the browser, with a login for each person.
- Your documents loaded so staff can ask questions about them.
- A walkthrough for your staff: what to ask, and how to check an answer.
- Every month after that: updates, backups, and someone to call when something breaks.
- Open WebUIChat screen with logins
- AnythingLLMAsk your documents
- Ollama or vLLMRuns the model
- Your GPU serverIn your office
What sits on the machine, from the screen staff use down to the graphics card.
For your IT personOllama or vLLM runs open models such as Llama, Qwen, Gemma or DeepSeek. Open WebUI is the chat screen. AnythingLLM or a custom document search answers from your files.
You see every number before you spend anything.
You buy the machine and own it outright. Our part is a one-time setup fee, then a monthly rate for updates, backups and support. Your written plan lists the machine, the setup fee and the monthly rate first.
From first call to running in under a week.
1
A short call
How many people, what kind of files, and what you want the AI to do.
2
Your written plan
The machine we would buy, the models it will run, the setup fee and the monthly rate. You decide with real numbers in front of you.
3
Setup, under a week
Once the machine is in your office, we install it, connect it, load your files and set up logins. It takes less than a week.
4
We look after it
The monthly rate covers updates, backups and support. When better open models come out, we tell you which ones are worth switching to.
An AI server sized to your office.
We tell you which machine fits the number of people and the work. You buy it, and you own it outright.
One person
A single desktop with a good graphics card. A way for one person to try private AI on their own work first.
A small office
One GPU workstation in a closet or server rack, shared by the team through the browser. Runs small and mid-size open models for a handful of people at once.
A larger team
A server with several graphics cards. Room for bigger models, longer documents, and many staff asking at the same time.
Private AI or business ChatGPT accounts?
If nothing sensitive goes into it, a business account may be all you need. We’d tell you that before selling you a server.
Business ChatGPT accounts are enough when
- The work is general writing with nothing sensitive in it.
- Only a few people use it now and then.
- You need the very strongest model for hard reasoning.
Your own server makes sense when
- Staff work with patient, client or financial files every day.
- You do not want another outside company holding that data.
- Many people would use it, so per-seat fees add up.
Smaller than the best cloud models
The open models that run on an office server are good at drafting, summarizing and answering from your files. For the hardest reasoning tasks, the largest cloud models are still ahead.
You buy the machine
A GPU workstation is a purchase in the thousands of dollars, and a multi-GPU server costs more. How big a model you can run is set mostly by the graphics card's memory.
Is local AI HIPAA compliant? It helps, but not by itself.
HHS says a cloud provider that stores or handles patient health information for you is a business associate and needs a business associate agreement, even if it cannot read the data (HHS guidance on cloud computing). Some cloud AI companies will sign one. Running the model on your own server means there is no outside AI company in the picture at all.
The rest still applies to the server like any other system holding patient data: access controls, audit logs, encryption, backups and your written policies (HHS Security Rule). We set up logins and keep the server inside your network. We do not certify anyone as HIPAA compliant.
General information, not legal or compliance advice. Check the current HHS guidance or ask a qualified professional.
Questions owners ask about private AI.
How long does setup take?
Less than a week once the machine is in your office. Before that, you get a written plan with the machine to buy and our prices, so you know what you are signing up for.
What does it cost?
Two parts: the machine, which you buy and own, and Wirestack's fee, which is a one-time setup fee plus a monthly rate for updates, backups and support. The written plan gives you all three numbers before you spend anything.
Is it as good as ChatGPT?
For everyday writing, summaries and questions about your own documents, a well-chosen open model does the job. For the hardest problems, the largest cloud models are still stronger. We tell you which of your tasks fit a local model before you buy hardware.
Does our data leave the building?
No. The model, the chat history and your documents sit on a server on your network. Prompts and files are not sent to an outside AI company. The server only goes online to download models and updates.
Who keeps it updated?
We do. Updates, backups and fixes are part of the monthly rate, the same way we run the other software we build.
We already use ChatGPT. Why switch?
You may not need to. If nothing sensitive goes into it, business accounts can be enough. Private AI is for the work you cannot put into an outside service.
Want AI wired into a process, not just a chat screen? See the processes we have automated, or read who we are.
The screens on this page are drawings, not a client’s system.
Free guide, PDF, 7 pages
Private AI for Business
A plain guide to running AI on a server you own: when it makes sense, what to plan for, and where HIPAA rules still apply.
Get a hardware plan before you spend anything.
Tell us how many people, what kind of files, and what you want the AI to do. You get a written plan with the machine to buy, the setup fee and the monthly rate. One of us will reply.
