Investors
The cloud where agents build and run software.
app.nz gives an agent everything it needs on one account: repos, models, GPUs, hosting, CI, search, and editors. We are a small technical team in New Zealand. The platform is live and billing customers, and we build it with the same agents we sell.
Why
Agents will write and run most software
Writing code is getting cheap. The work around it is not: repos, compute, models, deploys, review, and the bill. One agent session touches all of them, so they belong on one account.
20+
model providers behind one API
12
editors and studios live
$0.228/hr
GPU entry price, billed per second
1
credit balance for everything
Live today
What customers use now
Coding agents
Agents branch, edit, run tests, and open pull requests on hosted repos. Schedulers run them on cron; auto-agents run them on events.
One account
Repos, model gateway, GPU compute, hosting, CI, search, and editors bill to one prepaid balance under one key.
Model and GPU routing
The gateway fronts 20+ providers. Each request goes to the best model and the cheapest capable GPU, with automatic fallback.
Editors
Video, audio, image, vector, slides, sheets, docs, 3D, animation, and whiteboard. Every operation is a tool an agent can call.
The plan
Where this goes
Three things we are building toward. Each starts from something live today. The deep dive says what has shipped, what is next, and what is still research.
Infrastructure that optimizes itself
A learned router picks the model and the GPU per request, within about a millisecond, across every provider we can reach. Customers stop making infrastructure decisions.
Agents that run for months
Agents with durable memory, standing budgets, and escalation rules. They watch queues, patch dependencies, hold latency targets, and keep shipping.
Software that maintains itself
Systems that read their own telemetry, propose and test their own fixes, and merge through review gates a human set once.
Documents
Read more
10-year vision
Where software is going, what we are building for it, and the three stages we get there in.
Technical deep dive
Architecture, routing, GPU placement, security, economics, roadmap, and risks.
Security
Isolation, secret handling, spend limits, and audit trails.
Docs and API
The reference customers build against.
Talk to us
Try the product, then email the founder. Financials and cap table are shared under NDA.