Private AI for law firmsNZ & Australia
The time savings of AI — without putting privilege at risk.
A private assistant that answers questions and summarises from your firm's own documents — with sources — running on your own systems. No client or privileged information ever leaves the firm.
A preview of real, cited answers over a sample firm's own documents. Open the live assistant and ask your own questions — no sign-up.
The problem you already know
Client information can't go into public AI.
"Lawyers cannot safely enter confidential, sensitive or privileged client information into public AI chatbots like ChatGPT, or any other public tools." Doing so can waive privilege. — reflecting NZ & Australian Law Society guidance
But the time savings are real — and competitor firms are starting to capture them. The answer isn't to ban AI; it's to use it privately: on your documents, on your systems, inside the conduct rules.
How it works
Set up around your firm — not the other way round.
We point it at your documents
Your precedents, SOPs, policies and matter files — set up on your own systems.
Your team just asks
Plain-English questions; instant answers with the source shown, so every answer is verifiable.
Nothing leaves the firm
Private by design and set up to fit the conduct rules on confidentiality and privilege.
Prove it to yourself
Don't take my word for it — ask it something.
The assistant above is running live right now on a sample firm's handbook, hosted in Sydney. Open it and ask your own questions — every answer shows the document it came from, and if something isn't in the documents it says so instead of guessing. Ask it something off-topic and watch it decline.
No sign-up. Try: "Can I paste a client's affidavit into ChatGPT?" · "A new client wants to sue someone we acted for last year — is that a conflict?" · "Where do client retainer deposits go?" — or anything else. In your firm, the same assistant runs on your documents, in your own environment.
Who's behind it
One developer, hands-on.
I'm Daniel Olliver — a New Zealand developer who builds private AI systems. I've delivered production software handling sensitive data, including a patient-assessment platform for a UK healthcare provider — files with the same stakes as yours. I work with one firm at a time, hands-on, and start with a free demo on a few of your own (redacted) documents so you can see it before committing to anything. Everything I claim is verifiable — the demo above is live, and my wider work is public at danielolliver.com.
Start small
The AI Opportunity Audit — one week, fixed price, risk-free.
No open-ended consulting engagement, no strategy deck. One week, three concrete things, and if you don't find it worth the fee, you don't pay.
What you get
- A prioritised report: the top 3 places AI saves your firm time — ranked by impact, specific to how you practise.
- A working private assistant on a sample of your own documents — you see it answer, cited, not a slideshow about it.
- A clear build proposal with real numbers — what a full setup involves and costs, so you can decide either way.
Fixed. One week from a 30-minute kickoff.
Risk-free: not worth it, don't pay.
For the next few firms only.
The questions partners actually ask
Straight answers.
Is our client data used to train AI?
No. On the private-cloud setup, Microsoft's documented Azure OpenAI terms exclude your data from model training. On the on-premise setup, nothing leaves your building at all.
Where does our data live?
In your own Microsoft/Azure subscription — hosted in Sydney (Australia East) — or entirely on your own hardware. Never in a public chatbot. Firms with strict onshore-processing requirements have a dedicated-capacity option; we'd walk through it honestly.
What happens when it doesn't know?
It says so. Every answer cites the source document so staff can verify it — and questions the documents don't cover get a "that's not in the firm's documents" instead of a guess. Test that yourself.
Do we need new hardware or IT staff?
Private cloud: no hardware at all. On-premise: a modest office PC or small server. I set it up hands-on either way — your team just types questions.
What does it cost to start?
A $500 fixed-price, one-week audit (founding rate — normally $1,500), risk-free. The full build is quoted in the audit with real numbers, so you never commit blind.
We're a small firm with no tech team — is this for us?
That's exactly who it's built for: 2–20-lawyer firms. If AI genuinely wouldn't help your practice, the audit says that too — you're paying for an honest answer, not a sales pitch.
See it on your own documents.
A free, no-obligation 20-minute demo — on a few of your own redacted files, so you know exactly what it does for your firm.
Got here from my email? Just hit reply — it comes straight to me. Or try the live assistant first.