Blog
Notes on data, explainable AI, computer vision and local AI, written from the projects we build.
Local AI and the right-sized model for each task
You don't always need a huge model or a per-token bill. Small models on your own servers, step-by-step agents, and when a big model is worth it.
Read →Custom OCR pipelines: from reading a document to trusting the data
Why generic OCR falls short on real invoices and delivery notes, and how a pipeline with cross-checks and human review closes the gap.
Read →Explainable AI: why a model that's right isn't enough
What explainability means, how to read local and global explanations, and when not to trust a model even when it gets things right.
Read →