Workflows · Answers from your documents · Custom models · Video and media
Private AI, made to fit where and how you work.
We build it into the systems and documents you already use, and it runs on a machine in your building or on one of ours that serves only you. We work with you at every step, so you understand what it does and where it helps.
Start a conversationWhat private means here
It runs on a machine that serves only you: in your building, or on our hardware in the United States.
Nothing you send is passed to an outside AI provider.
Your material is never used to train a model for anyone else.
With a public AI service, others can look. With ZAITech, only you and we can.
With a public AI service. Your data and your questions leave your building and travel to the AI company, which can see what you send. Others stand along the way: authorities, who can ask the AI company for it; reviewers, on some plans; advertisers, because on some services what you send picks the ads you see. Who can see your data: you, the service and, on some services, others.
Who can see your data You, the service and, on some services, others
What we provide
Two ways to run it. The choice is yours.
With ZAITech, in your building. Your data never leaves your building: a question goes from your desk to your machine and the answer comes back. The machine serves only you. ZAITech builds it and looks after it, so ZAITech can see what is on it. In the drawing a knight, ZAITech, stands at the door, and the onlookers from the first picture stand outside the wall under a no sign, kept out. Who can see your data: you, and ZAITech.
Who can see your data You, and ZAITech
With ZAITech, on our hardware. Your data leaves your building locked: it is encrypted from your building to our machine. Nobody along the way can open it. It is opened only at one machine of ours, in the United States, that serves only you. ZAITech runs that machine and holds the key, so ZAITech can see what is on it. In the drawing your data travels in a locked chest past the onlookers from the first picture, who are kept out, and a knight, ZAITech, opens it with the key. Who can see your data: you, and ZAITech.
Who can see your data You, and ZAITech
Who can see your data depends on where it goes.
One stop for technology
Your one stop for technology, built around AI that stays private.
Many suppliers. One person with six technology needs around them. Each need has its own line to its own separate door, and the lines cross in a tangle.
One stop. The same person, calm, with the same six needs. Every line runs side by side to one door, where a knight, ZAITech, stands.
Bring us any technology need. We build it, run it, and answer for it.
AI · Software · Automation · Hardware · Media · Advice
You do not need to know where AI fits.
Finding that out is the first week of the work. We watch how the work moves through your team and change nothing. The workflows worth automating are usually not the ones on the list at the start.
Sort a week of ordinary work by two questions and the work worth automating is a short list; the rest stays with your people.
One picture of a week of work, sorted in three steps.
- A week of work, as it happens. Ten tasks, on the days we saw them. Monday: purchase requests, a supplier dispute. Tuesday: invoices, hiring interviews. Wednesday: policy questions, budget planning. Thursday: onboarding forms, a complaint. Friday: the monthly report, the annual audit.
- The same work, sorted by two questions. The ten tasks leave the week and land where the two questions put them. How often does it happen? Is it done the same way each time? Five tasks repeat and are done the same way: purchase requests, invoices, policy questions, onboarding forms and the monthly report. They are ringed. Hiring interviews and a complaint repeat but are different every time. The annual audit is done the same way but comes once a year. A supplier dispute and budget planning are rare and different every time.
- The first list. Three of the ringed tasks go on to the list. Purchase requests, built first. Invoices, next. Policy questions, next. Two more can follow. The rest stays with your people.
Ten tasks seen in one week, five in the ring, three on the first list.
Common problems we solve
Approvals
Which approvals could be handled without a person reading every form?
The monthly report
Can the monthly report be drafted from systems we already run?
Invoices and purchase requests
Can invoices and purchase requests be matched and queued for approval?
Policy questions
Can policy questions be answered from our own documents, source shown?
Where it pays
Where does AI pay for itself in our organisation, and where not yet?
What we build
Workflows and automation
The machine reads the forms, matches the invoices and drafts the monthly report from the systems you already run; a person signs off where it matters.
Answers from your own documents
Staff ask a policy or procedure question in plain words and get the answer with its source shown.
Custom models
A model trained on your material, so it knows your terms, your documents and your way of working.
Video and media
We produce and edit video, and generate pictures, music and other media alongside it.
What working with us is like
Made around your work.
A ready-made AI tool starts from the tool. We start from your work.
On site or remote.
We offer on-site service and deployment, or remote consultation. Either, as suits you.
A fixed scope and a fixed fee.
Both are agreed in writing before work starts.
We tell you what is not worth building.
A staff chatbot is a subscription, and cheaper. A feasibility check can end with us saying don’t build it.
You keep what we build.
A custom model is assigned to you in writing. A machine installed in your building is yours outright.
Where it runs is your choice.
In your building.
On a machine that is yours outright.
On our dedicated hardware.
On a machine in the United States that serves only you.
Every offer by name
Programs
- Pilot
Ninety days, one team: one workflow in production.
- Annual Program Recommended
Twelve months: five to six workflows in daily use.
Recommended because one workflow rarely pays for itself; five or six, found quarter by quarter, do.
- On Premises
The Annual Program on a machine in your building.
Inside the programs: Integration and automation, Software engineering.
Custom models
- Feasibility check
Three to five days: a written go or no-go.
- Proof of concept
Four to six weeks: a working model.
- Production model
Ten to sixteen weeks: trained, evaluated, assigned to you.
- Keeping it current
Drift monitoring, scheduled retraining.
Ongoing services
- Private AI endpoint
Hosted on our hardware. Single tenant, nothing retained.
- Installed at your site
A machine without a program: yours outright on delivery.
- Managed operation of your hardware
Monitoring, model updates, quarterly retuning.
- Monitoring only
Telemetry, alerting, a monthly health report.
- Advisory retainer
Vendor decisions, governance, roadmap.
FAQ
How are fees set?
Fixed and inclusive: no hourly rate underneath, no expenses line, no travel billing.
How long does it take?
In a program, the first workflow is live inside 30 days.
Can we stop?
Yes. After a Pilot, at any quarter boundary of an Annual Program, and after any stage of a custom model.
Talk to us
The first conversation is to find out whether this suits us both. It is free and needs none of your data.