About the imprint
Empire Publishing is an independent technical imprint. We publish field manuals for one of the hardest problems of this decade: as artificial intelligence grows more capable, the difficult part is no longer making it smart — it's making it trustworthy. Inspectable. Bounded. Measurable. Owned.
Every book we publish is written from the practitioner's chair. The frameworks are original and deployed; the case studies trace to real systems that were actually built, attacked, measured, and held — not thought experiments, and not the demo that worked once on stage. Where a number is real, it is real. Where it is illustrative, we say so. We would rather tell you what we caught lying than sell you a clean story.
That discipline is the brand. Our spine fits in one sentence: the future belongs not to the most capable AI, but to the AI you can trust.
Every framework is proven against a real system the author built and tried to break. If it never shipped, it doesn't go in the book.
Real numbers or honestly labeled ones. No overclaiming, no oracle promises — "above chance, not magic."
Reliable AI, Sovereign, and Private Intelligence each take the same thesis — trust over raw capability — from a different angle.
Reliable AI — the trilogy on engineering AI that survives contact with production: reasoning architecture, bounded autonomy, and the discipline of evaluation.
Sovereign — what happens when an AI agent can hold and spend real money with no human in the loop, and how to put the wallet behind a leash it cannot cut. Begins with Sovereign Machines.
Private Intelligence — the case that the real prize of local AI is not privacy but sovereignty: a mind on hardware you own, that you can see inside, steer, keep, and bound. Begins with Private Intelligence.
Stay in the loop
Get the free opening chapter of Sovereign Machines and a note the moment each new title goes live.
