Building safely with AI

Plain-language writing on shipping AI-built software you can rely on. Coherence, security, and keeping a product true to what you meant.

Guides & articles

What we are writing

Four pieces in progress, each one drawn from audits and builds we have run for clients. They cover the questions that come up before a launch.

Production

From vibe coding to production

An AI prototype gets you to a working screen in a day. Real users arrive with accounts, bad input, and load, and that is where the shortcuts underneath start to show.

Coming soon

What it covers

  • Where the shortcuts usually sit: auth, data access, error paths
  • The checks that make a demo safe for the first real accounts
  • What to fix before launch, and what can wait

Security

Securing a Supabase app

Row-level security policies that look right and are not, keys that ended up in the client bundle, and the access gaps that show up when the second user signs in.

Coming soon

Coherence

Keeping AI-built software true to intent

Drift is when the code still runs but no longer does what you asked for. What causes it, and how to catch it in review instead of in production.

Coming soon

Operations

No-silo automation

Connecting tools and teams so a change in one system does not quietly break the report someone else depends on.

Coming soon

Find out where your own app stands

Start with an audit. In a few days you'll know what's underneath, in plain language.