LIVE · Auckland, NZ · Zero-cost intelligence
Zero-Cost Intelligence · Auckland

Local AI for NZ business.
On your hardware.
Tailored to how you work.

Liveclaw deploys private AI that runs on your own hardware — models, memory and automation configured to your business, not rented per query from someone else's cloud. Once it's running, using it costs you nothing. That's zero-cost intelligence. Staged delivery, a fixed price per stage before we start, monitored after launch.

This is not software you run your business through. It's intelligence built around the way your business already works.

$0
Per-query cost once your AI is deployed
Yours
Runs on your hardware — data never leaves
24/7
Autonomous — working while you sleep
NZ-built
Auckland-based, on-site when it matters
CLUSTER READOUT
2.3B+
tokens through our agents
137
agent runs
5.9h
longest single agent run
21 days
current daily streak

Last updated 28 Sep 2026 · Source: our Cursor usage profile

  • GLM
  • nomic
  • Open WebUI
  • Hermes
  • DeepSeek
  • Kimi
  • OpenAI
  • OpenClaw
  • Goose
  • Whisper
  • Tavily
Sound familiar?

The four things every business asks us

Straight answers, not a technology pitch.

Q1
"I want to use AI but I don't know where to start."
You don't start with AI. You start with where your hours go. That's a 30-minute conversation about your week, not a technology pitch — and it's free. We'll tell you honestly if AI isn't the answer yet.
Q2
"I use Claude/ChatGPT as a chatbot, but I've no idea how to make it a tool for my business."
That's the gap almost every business is stuck in: talking to AI instead of AI doing work. We wire it into the jobs themselves — enquiries answered, bookings routed, reports drafted — so it works while nobody's watching a chat window.
Q3
"We hold sensitive data. How do we know AI won't compromise it?"
Because it never leaves your building. The models run on your own hardware — your customer data doesn't travel to anyone's cloud, including ours. We'll walk your IT person or advisor through the exact network path before anything is deployed.
Q4
"We don't want AI to become another agency-style cost that never ends."
Neither do we — that's the business we built against. Fixed price per stage, quoted before we start. Stop at any stage and keep everything built. And because it runs locally, there's no per-use meter — heavy use costs you nothing extra.
The questions worth asking

What most people don't think to ask AI

What does your business know that isn't written down?

We capture it, privately, so the business doesn't depend on one person's head.

Business Memory

What would you do if AI cost nothing per question?

Run it on everything, all day. Ours has processed over 2.3 billion tokens.

Always-on Reader

Which model is right for your job this month?

We test them every month on your real tasks, so you don't have to.

Model Fit Report

Where does your data legally have to live?

We'll show you exactly where it goes now, and what stays in your building.

Data Residency Check

We run every one of these on LIVECLAW first. You get what we already use.

Client work

Delivered and managed for NZ clients

Every engagement starts where the client needs it most. For Parked Up that was a website and booking automation. All sites below are live in production.

parkedup.co.nz
Live Parked Up — Event Venue, Victoria Park
Historic 1905 villa venue in Auckland's CBD. Full staged delivery: rescue form → complete site → booking-system integration.
AI applied: enquiries route straight into the venue's management dashboard, monitored weekly so none are silently lost. 11 enquiries in the first 6 days live.
Visit site
sharedbreath.nz
Live Shared Breath — Music Collective, Muriwai
Auckland songwriters recording from the west coast; debut EP "Sunset Muriwai". Minimalist, typography-led site.
AI applied: concept to deployed site in days through our AI build pipeline, at a budget a collective can afford.
Visit site
musicovermatter.co.nz
Live Music Over Matter — Independent Record Label
The Tāmaki Makaurau label behind Shared Breath. NZ On Air-supported releases, roster and philosophy.
AI applied: two related brands built and maintained through one AI pipeline — updated by automation, not agency retainers.
Visit site
What we deploy

AI that lives in your business, not in someone else's cloud

Every deployment runs the stack we operate in production ourselves — Ollama, Open WebUI and Qdrant — sized to your hardware and configured to the way your operation actually runs.

One machine or three. Sized to your business, fixed price, and yours to keep.

Get your hours back
Get back hours every week — handled by AI that runs on your own hardware, so heavy use costs you nothing.
  • Local AI deployment — private models on your own hardware, loaded with your documents, prices and processes.
  • Business automation & agents — enquiry routing, reporting, reconciliation, running autonomous and monitored, 24/7.
Tighter process, faster response
Fewer handoffs, faster follow-ups, fewer mistakes. Efficiency is where the revenue lift actually comes from — we just don't lead with the promise.
  • Websites & booking automation — a site that wins enquiries, dual-path delivery so nothing gets lost, weekly monitoring so we notice problems before you do.
A team doing the work that matters
Your people stop doing robot work. The robots are in the back room — yours, on your hardware — and your team decides what ships.
  • Tools that save time and money — an inbox reader that never misses a lead, a business memory your team can ask, overnight document audits. Built on your hardware, running every day.

Clients see revenue lift. We don't promise it — it arrives as a consequence of speed, accuracy and hours returned.

We test every frontier model as it ships, and every one that runs a client system earned its place. The full list is below. You get what's proven, not what's hyped.
Currently running client systems in production in Auckland.
De-risked by design

Three ways this is built to protect you, not us

Proven before purchase
We build and test your system on our own cluster first — you watch it working on samples of your real workflow before you commit to any hardware. Hardware is the last step, not the first.
Built for handover
Every system is documented and owned by you from day one. Take the reins whenever you're ready — next year or never. We design ourselves to be optional, not indispensable.
We run the frontier so you don't have to
We run the pro plans and test every frontier model as they ship — the latest reasoning gets applied to designing your system. Then it runs on sovereign local models in your building.
Frontier/cloud intelligence is design-time only — it works on your system's design, never on your customer data. Run-time is local, always.

Once it's running, using it costs you nothing.

How it works

A free conversation. A fixed-price plan you keep.

01
The Conversation
30 minutes on where your hours go. No pitch, no jargon. Honest answer if AI isn't the fit yet. Not your IT stack — where are you losing time? Which enquiries never come back?
Client risk: Free — the only free stage
02
The Scope Package (paid, fixed price — Audit + Discovery + Plan)
Baseline audit (hours, response times, error points) → discovery of what's worth automating → staged plan with named deliverables, measurables, and a fixed price per stage. The plan is yours to keep — take it to anyone.
Client risk: One known fee; you walk away owning the plan
03
The Proof Build
We build and test your system on OUR cluster with samples of your real workflow. You see it working before you buy any hardware.
Client risk: No hardware spend until you've watched it work
04
The Gate → Deploy
Your sign-off, then hardware + migration to your building. Every later stage has its own gate. Built for handover from day one — documented, owned by you.
Client risk: Stop at any gate; keep everything built
05
Measure & Improve
The loop runs: Understand → Streamline → Automate where it counts → Measure → Improve. Re-measured against your baseline at +30 days — the number, not a vibe. We reconcile your enquiry counts weekly and get alerted the moment something doesn't add up.
Client risk: No per-use meter, ever. Monitoring, if wanted, is a fixed line item you see coming

A free conversation. A fixed-price plan you keep. Proof on our hardware before you buy yours. Your sign-off at every stage. Then your own numbers tell you it worked.

Why work with us

The case for doing it this way

Your data stays yours
Your AI runs on your own hardware, in your building. Your strategy, clients, pricing and IP never leave the room, are never logged by a cloud provider, and never train someone else's model.
Zero-cost intelligence
Cloud AI charges you every time you think. Local AI has one deployment cost, then using it is free — ask it a thousand questions a day and the power bill barely notices. Fixed stage prices, agreed before we start.
Tailored, not generic
Not a chatbot with your logo on it. Models sized to your hardware, memory loaded with your documents and processes, automations built around the way your operation actually runs.
It runs itself — and we watch it
These aren't tools you have to remember to open. Systems run on schedule, deliver results, and alert us when something needs attention. The return is time.
The operators

Built by people running it live today

The infrastructure behind LIVECLAW is handling real projects right now — client websites, booking automation, music intelligence, venue operations. This is not a proof of concept.

Luke Thompson
Founder · Cluster Architect · Tāmaki Makaurau
Born and raised in Auckland. Built Red Bull Studios NZ — NZ's largest live music archive. Scaled FanPass from 11,000 to 60,000 subscribers at SKY TV. Now operating LIVECLAW's private AI cluster, running live projects every day, including the delivery pipeline behind every site on this page.
Pete Ohlson
Deployment · Client Delivery
Hands-on deployment specialist. Pete arrives on-site, installs, configures, walks your team through it, and makes sure everything is running before he leaves. Practical, direct, and accountable — the operator you want on the ground when it matters.
The stack

What we run, test and build with

We test the frontier, US and Chinese labs, open and closed, and run what earns its place. Not sure which model is right for your business? That's the job. We'll research it for you.

See the full stack

Running locally

  • Qwen 2.5 32B
  • Qwen 2.5 14B
  • Qwen 2.5 Coder 14B
  • Qwen3 32B
  • Qwen3 30B
  • GLM 4.7 Flash
  • Gemma 4 26B
  • Llama 3.2
  • nomic-embed

Systems we run

  • Ollama
  • Open WebUI
  • Qdrant
  • n8n
  • Hermes
  • Grok assistant
  • Cloudflare
  • Tailscale

Testing and evaluating

Open-weight

  • DeepSeek
  • Kimi K3
  • GLM (latest)
  • Nemotron 3

Frontier

  • Claude
  • GPT-6 Sol
  • GPT-6 Astra
  • Grok

Built with

  • OpenClaw
  • Goose
  • Whisper
  • ElevenLabs
  • Tavily
  • OpenRouter
  • Hugging Face
  • Cursor
  • Claude Code

Updated September 2026. Your data stays on local models unless you choose otherwise.

Let's have a conversation.

If we can add value to your business, we'll show you how — with your numbers. If we can't, we'll tell you that too.

Auckland-based · Response within 24 hours
Get in touch

Tell us what you're working with

We'll be straight with you about whether it's a fit — and what it would look like if it is. Or email us directly: [email protected]

No pitch decks. No sales calls. A direct conversation. How we handle your details
Message received. We'll come back to you within 24 hours.