Platform capabilities

You Can't Review Code
You Didn't Write.

That is the honest position most people are in now. The AI wrote it, it runs, the tests pass — and somewhere in there is a table without row-level security, a key in a config file, or a dependency with a known CVE. You would catch it if you had written it line by line. You didn't.

Every capability on this page exists to answer one question before your users do: is this actually safe to put on the internet? Six checks run on every deploy, in the same order, before a single visitor sees it.

What actually happens when you press deploy

Your code is scanned before anything is provisioned. Not after, not on a nightly job — before. Tables without row-level security, credentials sitting in files, dependencies carrying known CVEs, licence terms you did not intend to accept.

If it passes, the platform is configured for you and the app goes live on a real URL. If it fails, you get the finding and the file and the line, and nothing ships. That order is the entire product. A scan that runs after deployment is a report. A scan that runs before it is a gate.

Then it keeps watching. A health check runs against your production URL on a schedule. If the app goes dark, the previous working deployment is restored automatically and you are told what happened — you are not woken up to do it yourself at two in the morning.

Specifics, because “enterprise-grade” means nothing

AES-256-GCM

How stored credentials are encrypted at rest. Not hashed, not obfuscated — encrypted, with the key held outside the database.

Row-level security

Enforced on every table. Multi-tenant isolation is a database constraint here, not a WHERE clause someone has to remember.

Six checks, one order

Every deploy runs the same sequence. Not a configurable pipeline you can quietly disable when it is inconvenient.

Health checks on a schedule

Production URLs are probed continuously. Faster cadence on higher tiers — the exact interval is on the pricing page, and it matches what the scheduler actually runs.

Automatic rollback

A failed probe restores your previous working deployment. This is rollback, not restart: it puts back a deploy that worked.

Self-host export

Your code, your configuration, your database schema — exported and runnable elsewhere. The exit is built in, because a platform you cannot leave is a platform that can raise your price.

The thing nobody says out loud

The reason your app is still on localhost is not that the code is unfinished. It is that you cannot personally vouch for it, and putting your name on something you cannot vouch for is a specific kind of exposed that no amount of “it works on my machine” fixes.

You do not need to become a security engineer to ship. You need the check to have been run, and you need to be able to point at what it found.

What you get, and what it costs

Scanning, secrets vault, staged rollout, preview environments, roles and permissions, audit logs, automatic rollback and self-host export are included on every paid plan. What changes with tier is capacity — app installs, seats, workspaces, monitoring cadence — plus a few capabilities that only make sense higher up: bring-your-own AWS account on Business, white-label on Agency Pro.

Bring your own AI key on any plan, including Free. Most platforms in this category treat that as a premium feature. It is not one here, because your key means your provider bill and no meter of ours in between.

Flat monthly price. No per-request meter, no credits that expire, no invoice that arrives larger than you planned for because a page got popular.

Scan the repo you are already worried about

You do not have to take any of this on faith. Point ZipLoom at the project sitting on your machine right now and read what comes back.

Founding pricing is still open — see the offer →