Private, self-hosted and on-premise AI for Australian firms — the regulatory drivers, the honest trade-offs, and how to decide what keeps your client data in-house.
Private AI Setup - Private AI for your firm — set up so your data never leaves
You want AI working across your business, but your client data can't go to a public model. We set up a private AI assistant for you, end to end — a contractually-closed enterprise model or a fully self-hosted one, connected to your own documents, with Australian data residency and approve-before-it-acts controls. Fixed price, founder-led, handed over so you own it.
- Data never leaves your control
- Private
- Australian data residency
- On-shore
- Price, scoped up front
- Fixed
- Handed over to own
- Yours
What you get - A private AI assistant, set up end to end
One fixed-scope engagement. We configure it, connect your documents, train your team, and hand it over — you don't touch a setting.
- A private AI assistant, configured for your firm. A ChatGPT-style assistant for your team — running on a contractually-closed enterprise model (no training, no retention) or fully self-hosted, depending on how sensitive your data is.
- Your documents, answered — privately. We connect the assistant to your own files (retrieval) so staff get answers grounded in your matters, contracts, or records — without any of it leaving your environment.
- Australian data residency by default. Configured to keep data on-shore and aligned with the Privacy Act, the Australian Privacy Principles, and OAIC guidance on AI — so "where does our data go?" has a clean answer.
- Approve-before-it-acts controls. Guardrails and human-in-the-loop review so the AI assists and suggests, but a person signs off before anything is sent, filed, or relied on.
- A live walkthrough + handover. We don’t just hand over a configured tool — we train your team on the call and leave you with documentation, so you own it afterwards.
- Optional cloud fallback. If you want frontier-grade reasoning for the hardest tasks, we can add a routing layer that sends only non-sensitive work to a cloud model — your choice, your rules.
What it's actually for - The tasks your team hands it, day to day
“Private AI” is abstract until it's running your workday. Same tasks already eating your team's time — just none of it leaving your environment.
- Email, handled privately. Triage the inbox by priority and draft replies from your own pricing, policies and FAQs — the same grounded email assistant we build for clients, running on your private setup.
- Calendar & scheduling. Find times, handle reschedules, and prep you before a meeting with a one-line brief pulled from your own notes — in plain language, with no new app to learn.
- Meeting notes & call summaries. Transcribe and summarise client calls and internal meetings on your own infrastructure, so nothing said in a negotiation or consultation reaches a third-party API.
- Document Q&A over your own files. "What does clause 4 say?" "What did we agree with this client?" — answered instantly from your contracts, policies and records, not a web search.
- Client intake & front-desk. Answer routine client questions and booking enquiries around the clock, grounded in your actual policies — the same engine behind our AI receptionist, deployed privately.
- Invoices & data entry. Pull line items off invoices and receipts straight into your accounting system, without financial data ever touching a third party.
How it works - Scope, deploy, hand over
We keep it simple and fixed-price — no open-ended hourly consulting.
- 1 — Scoping call. We work out how sensitive your data is, which approach fits (contractually-closed enterprise vs self-hosted/on-prem), and what to connect. You get a fixed-price quote before anything is built.
- 2 — Build & configure. We deploy and configure the model, connect it to your documents, set Australian data residency and approve-before-it-acts controls, and test it on your real workflows.
- 3 — Walkthrough & handover. We train your team live, hand over documentation, and you own it. Optional capped maintenance retainer if you want it kept current.
What this is — and what it isn't.
This is a fixed-scope software setup and handover. It does not include 24×7 uptime guarantees, hardware procurement or hosting, air-gapped or classified environments, or work requiring formal IRAP/PROTECTED assessment. If that's what you need, we'll say so and point you to the right kind of provider. New to the topic? Start with our guide to private AI for Australian business.
- What is a Private AI Setup?
- It’s a fixed-scope, done-for-you engagement where we install and configure a private AI assistant for your business — one where your data never leaves your control — then train your team and hand it over. It spans the spectrum from a contractually-closed enterprise model (e.g. Claude with no-training, no-retention terms and Australian data residency) through to a fully self-hosted or on-premise model, chosen to fit how sensitive your data is.
- Do I need a fully on-premise / local model?
- Often not. For most firms, a contractually-closed enterprise model with Australian data residency and zero data retention is enough to satisfy privacy and confidentiality obligations. A fully self-hosted or on-premise model is for the most sensitive cases — privileged legal matters, certain health or financial data — where you want a hard guarantee that nothing transits a third party. We help you pick the right level rather than over-engineering it.
- Which industries is this for?
- Primarily regulated, sensitive-data professions: law firms, accounting and advisory practices, healthcare and allied health, financial services, and mortgage/finance broking — anyone who holds client data they can’t risk sending to a public AI tool.
- How much does it cost?
- It’s a fixed-price engagement scoped to your firm — typically from A$2,500 + GST for a setup-and-handover, depending on the model, the tools we connect, and how much of your document base we wire in. We scope and quote up front, so there are no surprises. Optional ongoing maintenance is available as a separate, capped retainer.
- What is out of scope?
- This is a software setup-and-handover service. It does not include 24×7 uptime guarantees, hardware procurement or hosting, air-gapped or classified environments, or anything requiring formal IRAP/PROTECTED assessment. If that’s what you need, we’ll tell you and point you to the right kind of provider.
- Is this Claude-only?
- No. We’re a Claude-first shop and Claude is excellent, but for a self-hosted setup we use the best open-weight model that fits your hardware and needs (Llama, Qwen, Mistral, and others). The point is private AI that fits your firm — not one vendor.
Further reading - Before you scope a private AI setup
How private AI works, the trade-offs, and where it fits.
After the Fable 5 ban, cloud-AI dependency is a real business risk. What local LLMs are, the honest trade-offs, and a pragmatic hybrid strategy.
The best open-weight models, how quantization works, real VRAM/RAM requirements by model size, and step-by-step setup with Ollama, LM Studio, and vLLM.
How AI consulting engagements are scoped, priced, and delivered — and the questions to ask before signing.
Let's build something great together.
Melbourne, Australia — serving founders worldwide. [email protected]