Private AI · Brisbane
Your own AI. In one box. In your building.
A ready-to-run AI server for your office. Fast enough for the whole team, private by design, and yours at one fixed price.
Summarise this 70-page contract and list the termination clauses.
Playback at measured single-user speed.
-
~220tok/s
Fast for one
Answers stream around 40× faster than you can read.
-
8at once
Fast for the team
Eight people at once, each still over 110 tokens/sec.
-
~70pages
Read in under 7 s
Long documents read in full before it replies.
-
1box
Nothing leaves
Plugs in on-site. Your data stays in the building.
The box
One box. Everything included.
- The server, set up and tested before it arrives
- Private AI models installed and tuned
- Search and Q&A across your documents
- On-site installation and staff handover
- Updates as better models come out
- Ongoing support options
Prefer the cloud or a mix? We do that too — ask us.
Performance
What one box does.
Measured in-house on a single box, October 2026. No cloud, no tricks, just the box on a desk.
Built for the whole office
Total output as more people use it at once, in tokens per second.
Reads long documents fast
Time to read a whole document before replying.
Read this before quoting us
- These are speed measurements, not answer-quality scores.
- The models think before they answer. The first word appears almost instantly; the first word of the actual answer typically takes about 0.4–1.8 seconds, depending on the task.
- Measured on one box configuration in October 2026. Real-world speed depends on your workload, and we confirm it as part of your quote.
The numbers as tables
| People | Tokens/sec (total) |
|---|---|
| 1 | 210 |
| 2 | 330 |
| 4 | 500 |
| 8 | 680 (110+ each) |
| Document | Read time |
|---|---|
| ~2 pages | 0.2 s |
| ~9 pages | 0.9 s |
| ~18 pages | 1.7 s |
| ~35 pages | 3.3 s |
| ~70 pages | 6.6 s |
Who it's for
For work that can't leave the office.
Built for firms whose documents are the business, and who'd rather not upload them anywhere.
-
Legal
Search and summarise matter files without uploading them anywhere.
-
Accounting & finance
Query years of working papers and draft client letters, all in-house.
-
Medical & allied health
Summarise referrals and notes without patient files leaving the practice.
-
Engineering & architecture
Find answers across specs, standards and past projects in seconds.
-
Government suppliers
Work through tenders and contracts on hardware you control.
Services
From model to running system.
One team for the whole job: deciding what to build, building it, and keeping it running afterwards.
-
01
On-Site AI
We build and deploy AI systems directly in your environment, so you own the stack and the data.
- Private language models on your own servers
- Search and Q&A across internal documents
- Hardware sizing, setup and handover
-
02
Hosting & Infrastructure
Web hosting, servers, and the plumbing that keeps everything running reliably.
- Websites and web apps, hosted and maintained
- Servers, backups and monitoring
- Networking and secure remote access
-
03
Consultancy
Clear, practical advice on what to build, where to run it, and how to keep it simple.
- Working out what's worth automating
- Choosing between on-site, cloud and hybrid
- Cost and hardware estimates before you commit
How we work
How a project runs.
No long discovery phase. You'll know what we're building, where it runs and what it costs before any work starts.
-
Talk
Tell us what you're trying to do. We'll say plainly whether AI is the right tool for it.
-
Scope
A written plan covering what we'll build, where it runs, and what it costs.
-
Build
We set it up on your hardware or ours, and test it with the people who'll use it.
-
Run
We keep it running, or hand it over with documentation so your team can.
We keep it simple, self-hosted, and under your control.
-
Your data stays where you put it
On-site by default. When something does need the cloud, you decide what goes there.
-
You own what we build
Open tooling, your hardware, your accounts. Nothing that only we can operate.
-
Plain answers
Simple systems are easier to run and cheaper to keep. We'll always suggest the simplest one that works.
FAQ
Fair questions.
Anything else, ask us. We'll give you a straight answer.
How is this different from ChatGPT?
It runs on hardware in your office. Your prompts and documents are processed on the box, not sent to a third party.
What's inside the box?
Hardware chosen and tuned for fast, private AI. We'll walk you through it in your quote.
How many people can use it at once?
We've tested eight at once, each still getting over 110 tokens a second. For bigger teams, we'll size it with you.
Can it read our documents and PDFs?
Yes. Search and Q&A across your internal documents is one of the main things it's set up to do.
What happens when better models come out?
We update the box.
How much does it cost?
One fixed price, quoted for your setup. Tell us a little about your team and we'll come back with a number.
Get in touch
Get a fixed-price quote.
Tell us what you'd like it to do. We'll come back with one price and a plain explanation.