Workshop
For people in the trade: the tools we build and how.
8 articles
AI · immagine generataA printed catalogue from our websites' data: what paper won't forgive
We generated an A5 catalogue from the same data interface behind our websites. Landscape photos, variable fonts and a forgotten placeholder reminded us that print plays by its own rules.
AI · video animatoA newsroom that checks itself (and knows when to stop)
How we write reader articles with AI: drafts only from the facts in the brief, four checks before publishing, and one article a week that goes out without human review only if every hard claim is confirmed.
AI · visione da renderAI Act, Article 50: what it means in practice for an estate agency
Chatbots that say they're bots, labelled retouched photos, disclosed AI articles: the AI Act's transparency rules, read by people who apply them every day. With the right dates and the open questions.
AI · video generatoAI-retouched listing photos: our notebook of mistakes
What generative models really do to photos of a home, the mistakes we paid for, and the transparency rules that came out of them: labels, originals, no simulations in disguise.
AI · immagine generataFifty-odd automations and one registry to know they're alive
Every process that runs on its own has a name, a runner, a proof of life and an off switch. Why a dashboard without proof lies, and how we found out.
AI · immagine generataThe risk isn't in the code: it's in agents that read other people's text
An internal security review showed us where the real danger lies: AI agents that mistake text written by others for orders. Here are the three rules we've set ourselves.
AI · immagine generataWe rebuilt our CRM with Claude, one table at a time
From a no-code Airtable system to our own app on Postgres, without ever switching the old one off. What worked, what broke, and the rules that came out of it.
AI · immagine generataWhen not to use AI: our matchmaker without a language model
Five questions, three homes and the reason for each: we built it without an LLM, using transparent scoring. Why plain code sometimes beats AI, and how to tell when.