What we actually think.
Short essays and frameworks from the Portial studio. No punditry, no hot takes. Just what we've learned from shipping AI systems that have to stay up on Monday morning.
The three places AI projects die — and how to avoid each one.
Most AI pilots fail for one of three reasons. None of them are "the model wasn't good enough". A diagnostic framework we use with every client before we commit to a Production Sprint.
Why a model in production is worth ten demos.
The gap between "we trained a model that works" and "a model is serving customers" is not technical. It's operational. A look at what production actually demands — and why so few consultancies will sign up for it.
The 92-day sprint: our delivery model, unpacked.
Week by week: what happens inside a Portial Production Sprint. Scope, milestones, the people involved, and what the client owns at the end.
Drift isn't a bug. It's a feature request.
Every model degrades. The question is whether you catch it before your users do. How we monitor, alert, and retrain — without creating a second full-time job for the data team.
The EU AI Act, in seven decisions.
How we translate the Act into seven concrete architectural decisions you make on day one of a project — before a single model is trained.
Want the essays when they land?
No newsletter rush. One email per essay, roughly monthly. Unsubscribe in one click.
Sign up below →