We open-sourced our investor room
The app.nz data room is now public pages: the 10-year vision, a full technical deep dive (architecture, ~1 ms routing, security, economics, benchmarks, roadmap, risks), all versioned in the repo.
Most startups keep two stories: the public one on the homepage and the real one in a data room behind an NDA. We think that split is a bug, so we deleted it. The app.nz investor room is now a set of public pages, versioned in the repo like any other code:
- /investors — the room itself: thesis, documents, and the long-term bets.
- /investors/vision — the 10-year vision: why this company should exist and why the opportunity is large.
- /investors/deep-dive — the technical deep dive: architecture, ~1 ms learned model routing, GPU placement across serverless and hosted pods on any provider, security, economics, benchmarks methodology, roadmap, and risks.
Why open source a data room
Three reasons.
Checkable beats polished. Every claim in the documents links to a live product surface — the gateway, the agent runtime, the editors, VisualBench. Diligence that starts from a running system is faster and more honest than diligence that reconstructs reality from a deck.
Documents rot; pages get maintained. An exported PDF is stale the day it is sent. These pages ship with the site, so updating the roadmap is a commit, and the diff is public like any other.
We sell openness. The platform's pitch to customers is replayable agent traces, public benchmarks methodology, and open code. The investor room is the same posture applied to ourselves.
What stays private
Customer-level financials, cohorts, and the cap table live in a private annex shared under NDA during a raise. Everything structural — pricing, mechanisms, margin logic, architecture, risks — is public.
The short version of the thesis
The marginal cost of writing software is collapsing. What stays scarce is the loop around the code: repos, models, GPUs, deploys, budgets, and review gates. app.nz exists to own that loop for a world where most software is written and operated by agents — with infrastructure that optimizes itself (per-request model and GPU routing inside a millisecond-scale budget), AI-native editors for every content type, and, over the decade, run-forever agents and self-organizing software.
If that sounds worth arguing with, good — the documents are written to be argued with. Read them, then email us.