Zombify turns AI model lifecycle risk into something developers can actually see, check and understand. The public experience is deliberately simple; behind it sits a human-reviewed registry, ingestion system and operational control layer built to keep the signal trustworthy.
Why it stays live
Zombify solves a real problem: model IDs are production dependencies, and providers can deprecate, rename, restrict or retire them underneath an application.
The public product gives that risk a clear, memorable interface. The deeper value is the system behind it: structured lifecycle data, observable ingestion, source confidence, human review and an admin layer that makes uncertain signals manageable.
That makes Zombify useful twice: as a live product people can use, and as working proof of how Pixelhatch connects product thinking, AI-assisted data operations and human control.
What was actually built
The strongest proof in Zombify is the operational layer behind the joke: a lifecycle registry with ingestion, auditability and human control.
The public hook is immediate utility: choose a provider and model ID, or surface IDs from a pasted .env file, then check whether the dependency is active, at risk, deprecated, zombie or retired.
components/ModelChecker.astro · api/check-model.ts The product grew beyond a one-off checker into a monitored model stack with magic-link access, lifecycle alerts and shareable status views — turning lifecycle data into something teams can actually use.
pages/dashboard.astro · pages/status/[token].astro The interesting part sits behind the public interface: pending, approved and rejected candidates; manual overrides; live-registry editing; lifecycle events; source confidence and explicit human review before uncertain signals become public truth.
pages/admin/registry-review.astro · migrations/003–006 Provider ingestion is observable rather than magical. Runs, row-level diffs, source status and health checks sit beside a newsroom workflow with draft, review, published and archived states, source classification and copyright-safe publishing guardrails.
api/admin-ingest-* · lib/newsroomIngest.ts · migrations/007–016 Product proof
Architecture
The system combines a public developer utility with server routes, a structured lifecycle database, authentication, ingestion observability and an editorial review layer.
A working product is stronger proof than a mockup.
The checker, Graveyard and Newsroom as a clear public product for understanding model lifecycle risk.
Back public status with source confidence, lifecycle events, ingest health and explicit human review.
The Registry Review control room as proof that a simple front end can sit on top of a serious operational system.
Zombify demonstrates how Pixelhatch can connect brand, product UX, data systems, AI-assisted ingestion and human operations.